Documente Academic
Documente Profesional
Documente Cultură
ibm.com/redbooks
International Technical Support Organization
November 2006
SG24-6788-02
Note: Before using this information and the product it supports, read the information in “Notices” on
page xxv.
This edition applies to features, microcode, GUI, and DS CLI as announced for the DS8000 in August 2006.
© Copyright International Business Machines Corporation 2005, 2006. All rights reserved.
Note to U.S. Government Users Restricted Rights -- Use, duplication or disclosure restricted by GSA ADP Schedule
Contract with IBM Corp.
Contents
Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii
Notices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxv
Trademarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxvi
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxvii
The team that wrote this redbook. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxvii
Special thanks to . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxix
Become a published author . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxx
Comments welcome. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxx
Part 1. Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Chapter 1. Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1 Introduction to Copy Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.1.1 Point-in-Time Copy (FlashCopy). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.1.2 Remote Mirror and Copy RMC (formerly Peer-to-Peer Remote Copy). . . . . . . . . . 5
Part 2. Interfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Part 3. FlashCopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
Contents v
12.11 Hardware requirements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
Contents vii
20.1 Synchronous and non synchronous data replication . . . . . . . . . . . . . . . . . . . . . . . . 244
20.1.1 Synchronous data replication and dependent writes . . . . . . . . . . . . . . . . . . . . 244
20.1.2 Asynchronous data replication and dependent writes. . . . . . . . . . . . . . . . . . . . 248
20.2 Basic concepts of Global Mirror . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
20.3 Set up a Global Mirror session . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
20.3.1 Simple configuration to start . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254
20.3.2 Establish connectivity to remote site . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254
20.3.3 Create Global Copy relationships . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
20.3.4 Introduce FlashCopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
20.3.5 Define the Global Mirror session. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
20.3.6 Populate the Global Mirror session with volumes . . . . . . . . . . . . . . . . . . . . . . . 258
20.3.7 Start the Global Mirror session . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
20.4 Consistency Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
20.4.1 Consistency Group formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260
20.4.2 Consistency Group parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
viii IBM System Storage DS8000 Series: Copy Services in Open Environments
24.1 Set up a Global Mirror environment using the DS CLI . . . . . . . . . . . . . . . . . . . . . . . 306
24.1.1 Preparing to work with the DS CLI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 306
24.1.2 Configuration used for the environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 306
24.1.3 Setup procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
24.1.4 Create Global Copy relationships - A to B volumes . . . . . . . . . . . . . . . . . . . . . 307
24.1.5 Create FlashCopy relationships - B to C volumes . . . . . . . . . . . . . . . . . . . . . . 308
24.1.6 Start Global Mirror. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
24.2 Remove a Global Mirror environment with the DS CLI . . . . . . . . . . . . . . . . . . . . . . . 315
24.2.1 End Global Mirror processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 316
24.2.2 Remove the A volumes from the Global Mirror session . . . . . . . . . . . . . . . . . . 317
24.2.3 Remove the Global Mirror session . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318
24.2.4 Terminate FlashCopy pairs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 319
24.2.5 Terminate Global Copy pairs and remove the paths. . . . . . . . . . . . . . . . . . . . . 319
24.3 Manage the Global Mirror environment with the DS CLI. . . . . . . . . . . . . . . . . . . . . . 320
24.3.1 Pause and resume Global Mirror Consistency Group formation. . . . . . . . . . . . 320
24.3.2 Change the Global Mirror tuning parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 323
24.3.3 Stop and start Global Mirror . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 324
24.3.4 Add and remove A volumes to the Global Mirror environment . . . . . . . . . . . . . 325
24.3.5 Add and remove an LSS to an existing Global Mirror environment . . . . . . . . . 327
24.3.6 Add and remove a subordinate disk subsystem . . . . . . . . . . . . . . . . . . . . . . . . 329
24.4 Recovery scenario after local site failure with the DS CLI. . . . . . . . . . . . . . . . . . . . . 330
24.4.1 Stop Global Mirror processing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 330
24.4.2 Perform Global Copy Failover from B to A . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331
24.4.3 Verify for valid Consistency Group state . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332
24.4.4 Reverse FlashCopy from B to C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 334
24.4.5 Re-establish the FlashCopy relationship from B to C . . . . . . . . . . . . . . . . . . . . 337
24.4.6 Restart the application at the remote site . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 338
24.5 Return to the local site . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 338
24.5.1 Create paths from B to A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339
24.5.2 Perform Global Copy Failback from B to A . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339
24.5.3 Query for the Global Copy first pass completion. . . . . . . . . . . . . . . . . . . . . . . . 341
24.5.4 Quiesce the application at the remote site . . . . . . . . . . . . . . . . . . . . . . . . . . . . 342
24.5.5 Query the Out Of Sync Tracks until it shows zero . . . . . . . . . . . . . . . . . . . . . . 342
24.5.6 Create paths from A to B if they do not exist. . . . . . . . . . . . . . . . . . . . . . . . . . . 342
24.5.7 Perform Global Copy Failover from A to B . . . . . . . . . . . . . . . . . . . . . . . . . . . . 343
24.5.8 Perform Global Copy Failback from A to B . . . . . . . . . . . . . . . . . . . . . . . . . . . . 344
24.5.9 Start Global Mirror. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 345
24.5.10 Start the application at the local site . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347
24.6 Practice disaster recovery readiness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347
24.6.1 Query the Global Mirror environment to look at the situation . . . . . . . . . . . . . . 348
24.6.2 Pause Global Mirror and check its completion . . . . . . . . . . . . . . . . . . . . . . . . . 348
24.6.3 Pause Global Copy pairs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 349
24.6.4 Perform Global Copy Failover from B to A . . . . . . . . . . . . . . . . . . . . . . . . . . . . 349
24.6.5 Create consistent data on B volumes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 350
24.6.6 Wait for the FlashCopy background copy to complete . . . . . . . . . . . . . . . . . . . 350
24.6.7 Re-establish the FlashCopy relationships. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 350
24.6.8 Take FlashCopy from B to D. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
24.6.9 Perform the disaster recovery testing using the D volume . . . . . . . . . . . . . . . . 352
24.6.10 Perform Global Copy Failback from A to B . . . . . . . . . . . . . . . . . . . . . . . . . . . 352
24.6.11 Wait for the Global Copy first pass to complete . . . . . . . . . . . . . . . . . . . . . . . 354
24.6.12 Resume Global Mirror . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354
24.7 DS Storage Manager GUI examples. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355
24.8 Setup a Global Mirror environment with the DS GUI. . . . . . . . . . . . . . . . . . . . . . . . . 355
Contents ix
24.8.1 Define paths . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355
24.8.2 Create Global Copy pairs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 359
24.8.3 Create FlashCopy relationships . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363
24.8.4 Create the Global Mirror session . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 367
24.9 Manage the Global Mirror environment with the DS GUI . . . . . . . . . . . . . . . . . . . . . 369
24.9.1 View settings and error information of the Global Mirror session . . . . . . . . . . . 370
24.9.2 View information of volumes in the Global Mirror session . . . . . . . . . . . . . . . . 371
24.9.3 Pause a Global Mirror session . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 372
24.9.4 Resume a Global Mirror session. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 372
24.9.5 Modify a Global Mirror session . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 373
Contents xi
31.2.2 Going from Global Mirror to Incremental Resync Metro/Global Mirror . . . . . . . 448
31.3 Failure at the local site scenario . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 454
31.3.1 Local site fails . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 455
31.3.2 Local site is back. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 457
31.3.3 Returning to the original configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 463
31.4 Failure at intermediate site scenario . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 472
31.4.1 Intermediate site failure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473
31.4.2 Intermediate site back up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 477
31.4.3 Re-synchronization at intermediate. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 480
31.4.4 Restore original configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 482
xii IBM System Storage DS8000 Series: Copy Services in Open Environments
34.3.2 Path creation using the DS GUI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 520
34.3.3 Establish logical paths between DS8000 and ESS 800 using DS CLI . . . . . . . 523
34.4 Managing Metro Mirror or Global Copy pairs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 526
34.4.1 Managing Metro Mirror or Global Copy pairs using the DS GUI . . . . . . . . . . . . 526
34.4.2 Managing Metro Mirror pairs using the DS CLI. . . . . . . . . . . . . . . . . . . . . . . . . 530
34.4.3 Managing Global Copy pairs using the DS CLI. . . . . . . . . . . . . . . . . . . . . . . . . 531
34.5 Managing ESS 800 Global Mirror . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 531
34.5.1 Managing Global Mirror pairs using the DS CLI . . . . . . . . . . . . . . . . . . . . . . . . 532
34.6 Managing ESS 800 FlashCopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 532
34.6.1 Creating an ESS 800 FlashCopy using the DS GUI . . . . . . . . . . . . . . . . . . . . . 533
34.6.2 Creating an ESS 800 FlashCopy using DS CLI . . . . . . . . . . . . . . . . . . . . . . . . 534
34.6.3 Creating a remote FlashCopy on an ESS 800 using DS CLI . . . . . . . . . . . . . . 535
Contents xiii
Chapter 36. IBM TotalStorage Rapid Data Recovery for UNIX and Windows . . . . . . 575
36.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 576
36.1.1 Solution highlights . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 576
36.2 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 576
36.3 Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 580
36.4 Additional information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 583
xiv IBM System Storage DS8000 Series: Copy Services in Open Environments
Solution benefits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 673
Planning and requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 674
Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 674
Implementation and usage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 675
Global Mirror for the entire disk space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 682
Solution description. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 682
Solution benefits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 683
Planning and requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 684
Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 684
Implementation and usage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 685
FlashCopy of IASP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 689
Solution description. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 689
Solution benefits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 691
Planning and requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 691
Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 691
Using Metro Mirror and FlashCopy of IASP in the same scenario . . . . . . . . . . . . . . . . 692
Implementation and usage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 692
FlashCopy of the entire disk space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 695
Solution description. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 695
Solution benefits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 696
Planning and requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 697
Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 697
Implementation and usage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 697
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 727
Contents xv
xvi IBM System Storage DS8000 Series: Copy Services in Open Environments
Figures
xviii IBM System Storage DS8000 Series: Copy Services in Open Environments
17-1 Paths panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
17-2 LSS Copy Options panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
17-3 Metro Mirror panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
17-4 Metro Mirror panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
19-1 DS8000 configuration in the Global Copy set up example . . . . . . . . . . . . . . . . . . . 214
19-2 Global Copy environment to set up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
19-3 The DS8000 environment for Global Copy offsite backup. . . . . . . . . . . . . . . . . . . . 224
19-4 Global Copy offsite backup scenario . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
19-5 Paths panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
19-6 Select source LSS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
19-7 Select target LSS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
19-8 Selecting source ports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
19-9 Select target I/O ports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
19-10 Select path options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
19-11 Verification panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
19-12 Paths defined . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
19-13 Metro Mirror panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
19-14 Volume Paring Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
19-15 Select source volumes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236
19-16 Select target volumes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236
19-17 Copy options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
19-18 Verification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
19-19 Global Copy volumes in copy pending status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
19-20 Out of sync tracks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
19-21 Selecting the action to convert to synchronous . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
19-22 Convert to synchronous confirmation panel. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
19-23 Metro Mirror panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
19-24 Metro Mirror panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
20-1 Synchronous data replication. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
20-2 Synchronous data replication and unnecessary recovery after local site failure . . . 245
20-3 Synchronous data replication, freeze, and restart without recovery required . . . . . 247
20-4 Asynchronous data replication. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
20-5 Global Copy sequence of data arrival not preserved at remote site . . . . . . . . . . . . 249
20-6 Global Copy non synchronous data replication involving multiple disk subsystems 250
20-7 Distributed application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
20-8 Global Mirror as a distributed application. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
20-9 Start with simple application environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254
20-10 Establish Global Copy connectivity between both sites. . . . . . . . . . . . . . . . . . . . . . 254
20-11 Establish Global Copy volume pair . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
20-12 Introducing FlashCopy in the Global MIrror solution . . . . . . . . . . . . . . . . . . . . . . . . 256
20-13 Define Global Mirror session . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
20-14 Add Global Copy source volumes to Global Mirror session. . . . . . . . . . . . . . . . . . . 258
20-15 Start Global Mirror . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
20-16 Formation of consistent set of volumes at the remote site. . . . . . . . . . . . . . . . . . . . 260
20-17 Consistency Group tuning parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
21-1 Global Mirror basic configuration with master and subordinate disk subsystems . . 266
21-2 Define Global Mirror session to all potentially involved storage disk subsystems . . 270
21-3 Decide for master disk subsystem and start Global Mirror session . . . . . . . . . . . . . 271
21-4 Global Mirror paths over FCP links between source storage disk subsystems . . . . 272
21-5 Dedicated and shared links . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273
21-6 Dedicated Global Mirror links and dedicated Global Copy links . . . . . . . . . . . . . . . 274
21-7 Normal Global Mirror operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
21-8 Production site fails . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276
Figures xix
21-9 Perform Global Copy Failover from B to A. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276
21-10 Check Consistency Group state . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
21-11 FlashCopy consistency group creation process . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
21-12 Set a consistent set of B volumes using the C volumes as source . . . . . . . . . . . . . 281
21-13 Restart applications at remote site. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
21-14 Failback operation from B to A in preparation for returning to local site . . . . . . . . . 283
21-15 Global Copy Failover from A to B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284
21-16 Global Copy failback from A to B and resync Global Copy volumes . . . . . . . . . . . . 284
21-17 Establish Global Mirror FlashCopy relationship between B and C . . . . . . . . . . . . . 285
21-18 Start Global Mirror session and resume application I/O at local site . . . . . . . . . . . . 285
22-1 Global Mirror main panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
22-2 Global Mirror pull-down list - no session selected . . . . . . . . . . . . . . . . . . . . . . . . . . 292
22-3 Global Mirror pull-down list - Session selected . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292
23-1 Global Copy with write hit at the remote site . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 296
23-2 Global Copy with overcommitted NVS at the remote site . . . . . . . . . . . . . . . . . . . . 297
23-3 Coordination time: How it impacts application write I/Os . . . . . . . . . . . . . . . . . . . . . 298
23-4 Remote disk subsystem with all ranks containing equal numbers of volumes . . . . 300
23-5 Remote disk subsystem with D volumes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
24-1 DS8000 configuration in the Global Mirror example . . . . . . . . . . . . . . . . . . . . . . . . 306
24-2 Start Global Mirror session with a subordinate DS8000#3 . . . . . . . . . . . . . . . . . . . 314
24-3 Global Mirror example before unplanned production site failure . . . . . . . . . . . . . . . 330
24-4 Site swap scenario after failoverpprc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331
24-5 Site swap scenario after the reverseflash command was issued . . . . . . . . . . . . . . 335
24-6 After the completion of the background copy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 336
24-7 After mkflash B to C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337
24-8 After the application start . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 338
24-9 Site swap scenario after Global Copy Failback from B to A . . . . . . . . . . . . . . . . . . 340
24-10 Site swap scenario after Global Copy Failover from A to B . . . . . . . . . . . . . . . . . . . 343
24-11 Site swap scenario after failbackpprc A to B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 344
24-12 Start Global Mirror . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 345
24-13 Start the application at the local site . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347
24-14 After mkflash B to D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
24-15 Perform Global Copy Failback A to B - test scenario. . . . . . . . . . . . . . . . . . . . . . . . 352
24-16 Global Copy paths creation - Launch the creation process . . . . . . . . . . . . . . . . . . . 356
24-17 Global Copy paths creation - step 1: select the source LSS . . . . . . . . . . . . . . . . . . 356
24-18 Global Copy paths creation - step 2: select the target LSS . . . . . . . . . . . . . . . . . . . 357
24-19 Global Copy paths creation - step 3: select the source I/O ports. . . . . . . . . . . . . . . 357
24-20 Global Copy paths creation - step 4: select the target I/O ports . . . . . . . . . . . . . . . 358
24-21 Global Copy paths creation - step 5: select the paths options. . . . . . . . . . . . . . . . . 358
24-22 Global Copy paths creation - step 6: verification . . . . . . . . . . . . . . . . . . . . . . . . . . . 359
24-23 Global Copy creation - Launch the creation process . . . . . . . . . . . . . . . . . . . . . . . . 359
24-24 Global Copy creation - step 1: choose volume pairing method . . . . . . . . . . . . . . . . 360
24-25 Global Copy creation - step 2: select the source volumes. . . . . . . . . . . . . . . . . . . . 360
24-26 Global Copy creation - Step 3: select the target volume 1. . . . . . . . . . . . . . . . . . . . 361
24-27 Global Copy creation - Step 3: select the target volume 2. . . . . . . . . . . . . . . . . . . . 361
24-28 Global Copy creation - Step 4: select the copy options . . . . . . . . . . . . . . . . . . . . . . 362
24-29 Global Copy creation - Step 5: verification. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363
24-30 FlashCopy creation - Launch the creation process . . . . . . . . . . . . . . . . . . . . . . . . . 364
24-31 FlashCopy creation - Step 1: Select the relationship type . . . . . . . . . . . . . . . . . . . . 364
24-32 FlashCopy creation - Step 2: select the source volumes. . . . . . . . . . . . . . . . . . . . . 365
24-33 FlashCopy creation - Step 3: select the target volumes . . . . . . . . . . . . . . . . . . . . . 365
24-34 FlashCopy creation - Step 4: select the common option . . . . . . . . . . . . . . . . . . . . . 366
24-35 FlashCopy creation - Step 5: verification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 366
Figures xxi
34-8 Select source I/O ports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 522
34-9 Select target I/O ports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 522
34-10 Path panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 523
34-11 Using ESS 800 Specialist GUI to display the ESS 800 WWNN . . . . . . . . . . . . . . . 524
34-12 Metro Mirror . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 527
34-13 Pairing method. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 527
34-14 Select source volume. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 528
34-15 First Metro Mirror target volume. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 528
34-16 Second Metro Mirror target volume . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 529
34-17 Copy options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 529
34-18 Managing Metro Mirror pairs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 530
34-19 Creating ESS 800 FlashCopy using DS GUI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 533
34-20 Selecting ESS 800 source volumes for FlashCopy . . . . . . . . . . . . . . . . . . . . . . . . . 534
35-1 TPC software products suite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 540
35-2 Management of Copy Services Functions supported by TPC for Replication . . . . . 542
35-3 Failover/failback terminology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 544
35-4 TPC for Replication - Metro Mirror copy sets. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 546
35-5 TPC for Replication - Global Mirror copy sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 547
35-6 TPC for Replication - session concept. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 548
35-7 TPC for Replication - four sessions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 552
35-8 TPC for Replication system overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 553
35-9 Replication Manager server connectivity to DS6000 and DS8000 . . . . . . . . . . . . . 554
35-10 Select Configure network ports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 555
35-11 TPC for Replication server freeze . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 558
35-12 LSS heartbeat triggers freeze when connectivity to server fails . . . . . . . . . . . . . . . 559
35-13 Windows 2003 Software level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 560
35-14 GUI URL determined during Replication Manager install time . . . . . . . . . . . . . . . . 561
35-15 TPC for Replication GUI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 562
35-16 Launch the Replication Manager GUI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 563
35-17 Health Overview panel. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 564
35-18 Sessions overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 565
35-19 About to create copy sets to session . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 566
35-20 Actions against a session . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 566
35-21 GUI basic layout. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 567
35-22 Select storage subsystem and select View/Modify Details action . . . . . . . . . . . . . . 568
35-23 Storage subsystem details. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 568
35-24 Path overview panel. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 569
35-25 Path overview of a DS8000 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 570
35-26 Replication Management Servers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 570
35-27 Advanced tools option . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 571
35-28 Console log . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 572
36-1 Sample update sequence in a database . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 577
36-2 Sample Rolling Disaster. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 578
36-3 Dependent write protection of database integrity. . . . . . . . . . . . . . . . . . . . . . . . . . . 579
36-4 eRCMF protection tier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 580
36-5 eRCMF architecture: 2-site configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 581
36-6 eRCMF architecture: 3-site configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 581
36-7 eRCMF issues freeze command in a disaster recovery scenario . . . . . . . . . . . . . . 582
37-1 TotalStorage Continuous Availability for Windows protection tier . . . . . . . . . . . . . . 588
37-2 Architecture of the solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 589
A-1 lspv output before recreating the volume group. . . . . . . . . . . . . . . . . . . . . . . . . . . . 595
A-2 lspv output after recreating the volume group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 595
A-3 Target file system stanza . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 596
xxii IBM System Storage DS8000 Series: Copy Services in Open Environments
A-4 Remote Mirror and Copy phantom disks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 597
A-5 Original time stamp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 597
A-6 Update source time stamp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 598
A-7 Update secondary server ODM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 598
A-8 Microsoft VSS architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 602
A-9 Microsoft VSS with DS8000 FlashCopy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 604
A-10 Microsoft VDS architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 605
A-11 QLogic SANsurfer VDS Manager startup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 607
A-12 Microsoft VDS and VSS architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 608
A-13 Microsoft VDS and VSS with DS8000 storage subsystem . . . . . . . . . . . . . . . . . . . 609
A-14 Virtual machine source disks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 618
A-15 Virtual disk types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 618
A-16 FlashCopy within the same virtual machine. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 619
A-17 FlashCopy within the same virtual machine - After FlashCopy . . . . . . . . . . . . . . . . 620
A-18 FlashCopy between virtual machines - Source virtual machine . . . . . . . . . . . . . . . 621
A-19 FlashCopy between virtual machines - Target virtual machine . . . . . . . . . . . . . . . . 622
A-20 FlashCopy between virtual machines - Target virtual machine after FlashCopy . . . 623
A-21 FlashCopy between ESX Servers - Source virtual machine . . . . . . . . . . . . . . . . . . 624
A-22 FlashCopy between ESX Servers - Target virtual machine. . . . . . . . . . . . . . . . . . . 625
A-23 FlashCopy between ESX Servers - Target virtual machine after FlashCopy . . . . . 626
B-1 System i5 partitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 631
B-2 Single level storage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 632
B-3 Input Output Processors (IOPs). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 633
B-4 Elements of System i5 Cluster. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 634
B-5 IASPs in System i5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 636
B-6 Metro mirror of IASP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 638
B-7 Functioning of the System i Copy Services Toolkit . . . . . . . . . . . . . . . . . . . . . . . . . 642
B-8 Checking status of IASP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 643
B-9 wrkcfgsts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 644
B-10 Status of IASP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 644
B-11 dspessdta. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 645
B-12 IASP name. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 646
B-13 Display Toolkit options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 646
B-14 chkpprc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 647
B-15 Checking PPRC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 648
B-16 chkpprc completed. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 649
B-17 viewlog . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 650
B-18 Toolkit log. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 651
B-19 Viewprof . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 651
B-20 DS CLI profiles on i5/OS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 652
B-21 Display DS CLI profile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 652
B-22 Select DS CLI script on i5/OS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 653
B-23 DS CLI script on i5/OS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 654
B-24 swpprc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 654
B-25 Parameter scheduled. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 655
B-26 Messages about progress-2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 656
B-27 Log of SWPPRC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 657
B-28 Metro Mirror copy if IASP after switchover. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 657
B-29 Original IASP after switchover . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 658
B-30 dspessdta after switchover . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 658
B-31 swpprcback to production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 659
B-32 Toolkit log after successful switchback . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 660
B-33 Status of IASP on production after switchback . . . . . . . . . . . . . . . . . . . . . . . . . . . . 661
Figures xxiii
B-34 Toolkit status after switchback to production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 661
B-35 swpprc*unscheduled . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 662
B-36 Message Waiting for reply . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 663
B-37 Display operator message . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 663
B-38 Confirming Unscheduled switch by replying the message. . . . . . . . . . . . . . . . . . . . 664
B-39 dspessdta after unscheduled switch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 665
B-40 SWPPRC *COMPLETE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 665
B-41 Waiting for reply to operator message . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 666
B-42 Confirm replicate from remote site to local site . . . . . . . . . . . . . . . . . . . . . . . . . . . . 666
B-43 Status after switch is completed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 667
B-44 Global Mirror of IASP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 669
B-45 falovrgmir . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 671
B-46 Metro Mirror of entire disk space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 673
B-47 Tagged loadsource . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 676
B-48 External loadsource . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 676
B-49 Properties of loadsource . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 677
B-50 IOP for Boot from SAN. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 677
B-51 LSPPRC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 678
B-52 Script for suspend Metro mirror . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 678
B-53 Performing script for suspend Metro Mirror . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 679
B-54 Script for failover of Metro Mirror . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 679
B-55 Performing script for Failover of Metro Mirror . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 679
B-56 Status of primary volumes after suspend and failover . . . . . . . . . . . . . . . . . . . . . . . 679
B-57 Status of secondary volumes after suspend and failover . . . . . . . . . . . . . . . . . . . . 680
B-58 IPL partition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 680
B-59 Script for failback of Metro Mirror. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 681
B-60 Performing script for failback of Metro Mirror. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 681
B-61 Pause Metro Mirror or secondary DS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 681
B-62 Fail over to the primary DS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 681
B-63 Failback on primary DS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 682
B-64 Global Mirror of entire disk space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 683
B-65 Pause GM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 686
B-66 Pause Global Copy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 686
B-67 Failover Global Copy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 686
B-68 Failback Global Copy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 687
B-69 Fail over to local site . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 687
B-70 Failback Global Copy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 687
B-71 Resume Global Mirror . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 687
B-72 Failover Global Copy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 688
B-73 LsFlash at GM unplanned outages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 689
B-74 Reverse FlashCopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 689
B-75 FlashCopy of an IASP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 690
B-76 PRPIASPBCK . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 693
B-77 Executing the command rsmiaspbck . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 694
B-78 FlashCopy of entire System i5 disk space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 696
37-3 Apply Activation Codes panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 712
E-1 ESS Copy Services GUI tasks panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 717
E-2 ESS task information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 717
E-3 ESS GUI FlashCopy window . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 718
xxiv IBM System Storage DS8000 Series: Copy Services in Open Environments
Notices
This information was developed for products and services offered in the U.S.A.
IBM may not offer the products, services, or features discussed in this document in other countries. Consult
your local IBM representative for information about the products and services currently available in your area.
Any reference to an IBM product, program, or service is not intended to state or imply that only that IBM
product, program, or service may be used. Any functionally equivalent product, program, or service that does
not infringe any IBM intellectual property right may be used instead. However, it is the user's responsibility to
evaluate and verify the operation of any non-IBM product, program, or service.
IBM may have patents or pending patent applications covering subject matter described in this document. The
furnishing of this document does not give you any license to these patents. You can send license inquiries, in
writing, to:
IBM Director of Licensing, IBM Corporation, North Castle Drive Armonk, NY 10504-1785 U.S.A.
The following paragraph does not apply to the United Kingdom or any other country where such provisions
are inconsistent with local law: INTERNATIONAL BUSINESS MACHINES CORPORATION PROVIDES THIS
PUBLICATION "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESS OR IMPLIED,
INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF NON-INFRINGEMENT,
MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Some states do not allow disclaimer of
express or implied warranties in certain transactions, therefore, this statement may not apply to you.
This information could include technical inaccuracies or typographical errors. Changes are periodically made
to the information herein; these changes will be incorporated in new editions of the publication. IBM may make
improvements and changes in the product(s) and the program(s) described in this publication at any time
without notice.
Any references in this information to non-IBM Web sites are provided for convenience only and do not in any
manner serve as an endorsement of those Web sites. The materials at those Web sites are not part of the
materials for this IBM product and use of those Web sites is at your own risk.
IBM may use or distribute any of the information you supply in any way it believes appropriate without
incurring any obligation to you.
Information concerning non-IBM products was obtained from the suppliers of those products, their published
announcements or other publicly available sources. IBM has not tested those products and cannot confirm the
accuracy of performance, compatibility or any other claims related to non-IBM products. Questions on the
capabilities of non-IBM products should be addressed to the suppliers of those products.
This information contains examples of data and reports used in daily business operations. To illustrate them
as completely as possible, the examples include the names of individuals, companies, brands, and products.
All of these names are fictitious and any similarity to the names and addresses used by an actual business
enterprise is entirely coincidental.
COPYRIGHT LICENSE:
This information contains sample application programs in source language, which illustrates programming
techniques on various operating platforms. You may copy, modify, and distribute these sample programs in
any form without payment to IBM, for the purposes of developing, using, marketing or distributing application
programs conforming to the application programming interface for the operating platform for which the sample
programs are written. These examples have not been thoroughly tested under all conditions. IBM, therefore,
cannot guarantee or imply reliability, serviceability, or function of these programs. You may copy, modify, and
distribute these sample programs in any form without payment to IBM for the purposes of developing, using,
marketing, or distributing application programs conforming to IBM's application programming interfaces.
SAP, and SAP logos are trademarks or registered trademarks of SAP AG in Germany and in several other
countries.
Java, JDK, Solaris, StorageTek, Sun, and all Java-based trademarks are trademarks of Sun Microsystems,
Inc. in the United States, other countries, or both.
Internet Explorer, Microsoft, Windows NT, Windows Server, Windows, and the Windows logo are trademarks
of Microsoft Corporation in the United States, other countries, or both.
Intel, Pentium, Intel logo, Intel Inside logo, and Intel Centrino logo are trademarks or registered trademarks of
Intel Corporation or its subsidiaries in the United States, other countries, or both.
UNIX is a registered trademark of The Open Group in the United States and other countries.
Linux is a trademark of Linus Torvalds in the United States, other countries, or both.
Other company, product, and service names may be trademarks or service marks of others.
xxvi IBM System Storage DS8000 Series: Copy Services in Open Environments
Preface
This IBM® Redbook will help you install, tailor, and configure Copy Services for open systems
environments on the IBM System Storage™ DS8000™. It should be read in conjunction with
The IBM TotalStorage DS8000 Series: Implementation, SG24-6786.
This book gives the details necessary to configure each of the Copy Services functions in an
open systems environment. There is a companion book that supports the configuration of the
Copy Services functions in a z/OS® environment, The IBM TotalStorage DS8000 Series:
Copy Services with IBM Eserver zSeries, SG24-6787.
This book will help you design/create a new Copy Services installation or migrate from an
existing installation, as well as address design functions and changes in terminology from
other IBM Copy Services products. We include hints and tips to maximize the effective
installation.
Bert Dufrasne is a Certified Consulting IT Specialist and Project Leader for IBM
TotalStorage® products at the International Technical Support Organization, San Jose
Center. He has worked at IBM in various IT areas. Before joining the ITSO, he worked for IBM
Global Services as an Application Architect. He holds a degree in Electrical Engineering.
Gustavo Castets is a Certified IBM IT Specialist and Project Leader working for the IBM
International Technical Support Organization, San Jose Center. While in San Jose, from
2001 to 2004, Gustavo co-authored 12 IBM Redbooks™ and taught IBM classes worldwide in
the area of disk storage systems. Before joining the ITSO, Gustavo was based in Buenos
Aires where he worked in different technical sales support positions for more than 22 years.
Today, in addition to his work with the ITSO, Gustavo works for IBM Global Delivery in
Argentina, as a Storage Specialist supporting accounts from the United States and Europe.
Stephen Baird is an IT Specialist with IBM Global Services. He joined IBM in 1999, working
in open systems server performance management and capacity planning. Since 2002, he has
worked in Storage Area Network and disk storage subsystem support and has gained
experience with Brocade, Cisco, and McData Fibre Channel switches and directors as well as
DS8000, DS4000™, and ESS series hardware. He holds a degree in Mechanical
Engineering from the Massachusetts Institute of Technology, in Cambridge, MA.
Denise Brown is a Software Engineer at the IBM Open Systems Level Test Lab in Tucson,
Arizona. She has been working with IBM Storage for the past four years, with experience in
Jana Jamsek is an IT Specialist in IBM Slovenia. She works in Storage Advanced Technical
Support for Europe as a specialist for IBM Storage Systems and i5/OS® systems. Jana has
eight years of experience in the System i™ and AS/400® area, and six years experience in
Storage. She holds a Master’s degree in computer science and a degree in mathematics from
University of Ljubljana, Slovenia. She has authored serveral IBM Redbooks and IBM
Redpapers on iSeries™ and TotalStorage.
Peter Klee is an IT Specialist with ATS EMEA, located in Mainz, Germany. He has 10 years
of experience in data centers. He worked for a large bank in Germany and was responsible
for the architecture and the implementation of the disk storage environment using EMC
Symmetrix, HDS Lightning, and ESS Model 800. He joined IBM in 2003, where he was
working for Strategic Outsourcing. Since June 2004 he is working in the ATS System Storage
team in Mainz. His main focus is Copy Services in the open systems environment with
DS8000 and DS6000, especially Global Mirror and Metro/Global Mirror.
Markus Oscheka is an IT Specialist for Proof of Concepts and Benchmarks at the ATS
Customer Solutions team in Mainz, Germany. His areas of expertise include setup and
demonstration of IBM System Storage products and solutions in various environments
including AIX®, Linux®, Windows®, HP-UX, and Solaris™. He has worked at IBM for five
years. He has performed many Proof of Concepts with Copy Services on DS6000/DS8000,
as well as Performance-Benchmarks with DS4000/DS6000/DS8000.
Ying Thia is an Advisory IT Specialist based in IBM Singapore, providing storage technical
support. She has 14 years of experience in the S/390 and storage environment. Before
joining the IBM Singapore Storage team, she worked in IBM Global Services where her
responsibilities included technical support and services delivery. Her areas of expertise
include IBM high-end disk and tape storage subsystems and disaster recovery solutions
using the capabilities and features of IBM storage products. She co-authored a previous IBM
Redbook and workshop for zSeries® Copy Services.
Robert Tondini is a Senior IT Specialist based in IBM Australia, providing storage technical
support. He has 12 years of experience in the S/390 and storage environment. His areas of
expertise include IBM high-end disk and tape storage subsystems and disaster recovery
solutions using the capabilities and features of IBM storage products. He co-authored several
IBM Redbooks and a workshop for zSeries copy services.
xxviii IBM System Storage DS8000 Series: Copy Services in Open Environments
The team: Gustavo, Robert, Wenzel, Jana, Peter, Markus, Denise, Werner, Ying, Stephen, Bertrand
Special thanks to
John Bynum and Robert Moon - Techincal Support Marketing Leads
IBM US
We want to thank Michael Eggloff and Peter Klee for hosting us at the European Storage
Competency Center in Mainz, Germany. They were able to supply us with the needed
hardware, conference room, and all of the many things needed to run a successful residency.
Günter Schmitt, Uwe Schweikhard, Edgar Strubel (ATS - IBM Mainz) for their help in
reserving and preparing the equipment we used.
Monika Baier, Susanne Balzer, Marion Barlen, Marion Hartmann, Andrea Witkowski, Gertrud
Kramer - IBM Mainz, for their help with administrative tasks.
Many thanks to the authors of the previous edition of this IBM Redbook:
Peter Kimmel, Jukka Myyrlainen, Lu Nguyen, Gero Schmidt, Shin Takata, Anthony
Vandewerdt, Bjoern Wesselbaum
Selwyn Dickey, Timothy Klubertanz, Vess Natchev, James McCord and Chuck Stupca
IBM Rochester - System i Client Technology Center
Bob Bartfai, Jennifer Mason, James Davison, Steve Van Gundy, Richard Ripberger, Bill
Raywinkle, Christopher o’Toole, Jim Sedgwick, Henry Sautter, Craig Gordon, Rosemary
McCutchen,, Lee La Frese, Alan McClure, Rachel Mikolajewski, Gail Spear, Leann Vaterlaus,
David V Valverde, Sonny Williams
Brian Sherman
IBM Canada
Nick Clayton
IBM UK
Preface xxix
Many thanks to Emma Jacobs and Julie Czubik.
Your efforts will help increase product acceptance and customer satisfaction. As a bonus,
you'll develop a network of contacts in IBM development labs, and increase your productivity
and marketability.
Find out more about the residency program, browse the residency index, and apply online at:
ibm.com/redbooks/residencies.html
Comments welcome
Your comments are important to us!
We want our Redbooks to be as helpful as possible. Send us your comments about this or
other Redbooks in one of the following ways:
Use the online Contact us review IBM Redbook form found at:
ibm.com/redbooks
Send your comments in an email to:
redbook@us.ibm.com
Mail your comments to:
IBM Corporation, International Technical Support Organization
Dept. HYTD Mail Station P099
2455 South Road
Poughkeepsie, NY 12601-5400
xxx IBM System Storage DS8000 Series: Copy Services in Open Environments
Summary of changes
This section describes the technical changes made in this edition of the book and in previous
editions. This edition may also include minor corrections and editorial changes that are not
identified.
Summary of Changes
for SG24-6788-02
for IBM System Storage DS8000 Series: Copy Services in Open Environments
as created or updated on November 28, 2006.
New information
3-site Metro/Global Mirror with Incremental Resync
IBM TotalStorage Productivity Center for Replication
Host operating systems considerations for System i
New DS8000 series Turbo models: 93x/9B2
Changed information
Updated Copy Services examples
Updated licensing information
Part 1 Overview
In this part we describe the various Advanced Copy Services offerings for the DS8000 series
and how they relate to previous Copy Services offerings available on the Enterprise Storage
Server® (ESS).
This part will also show how the existing Copy Services functions from the ESS can coexist
with the DS8000 series Copy Services. Similarly, we discuss their use with the DS6000
series Copy Services.
With the announcement of the DS8000 series and Advanced Copy Services for z/OS on the
DS8000 series, we introduced some new terminology. We also introduced these terms for
Advanced Copy Services Version 2 for the ESS. These are detailed in the following IBM
Redbook, The IBM TotalStorage DS8000 Series: Implementation, SG24-6786, and are
summarized in Table 1.
z/OS Global Mirror z/OS Global Mirror Extended Remote Copy (XRC)
Metro/Global Mirror for zSeries z/OS Metro/Global Mirror 3-site solution using Sync
PPRC and XRC
It should be noted:
All DS8000 installations require at least an Operating Equipment License (OEL) key to
operate. However, Copy Services functions can only be used where the appropriate
License Key is installed.
The key Copy Services license types are Operating Equipment License (OEL), Remote
Mirror and Copy or PPRC (RMC), and Point-in-Time Copy or FlashCopy (PTC). For
DS8000 there is also Remote mirror for z/OS (RMZ).
The DS CLI replaces both the ESS CLI and the ESS Copy Services CLI.
The DS CLI can also be used for ESS 800 Copy Services, but not ESS configuration. For
ESS configuration you need to continue to use the ESS Web Specialist or ESS CLI.
The DS CLI can invoke Remote Mirror relationships with the ESS 800, if the ESS 800 is at
code level LIC 2.4.3.65 or later.
The DS CLI can be used in reusable scripts, and provides a consistent interface for
current and planned IBM System Storage products.
The DS CLI now invokes a Copy Services function directly rather than invoking a saved
task.
The DS GUI can only be used for one-time execution of a Copy Services operation; it
cannot save tasks.
Chapter 1. Introduction
This chapter provides a brief summary of the various Copy Services functions available on
the DS8000 series. These services are very similar to the existing Copy Services for the IBM
Enterprise Storage Server, and some models of the ESS are interoperable with the DS8000
as well as the DS6000.
With the Copy Services functions, for example, you can create backup data with little or no
disruption to your application, and you can back up your application data to the remote site for
disaster recovery.
Copy Services run on the DS8000 Storage Unit and support open systems and zSeries
environments. These functions are supported also on the previous generation of storage
systems, the IBM TotalStorage Enterprise Storage Server (ESS).
Many design characteristics of the DS8000 and the data copying and mirroring capabilities of
the Copy Services features contribute to the protection of your data, 24 hours a day and 7
days a week (24x7). The licensed features included in Copy Services are as follows:
FlashCopy, which is a Point-in-Time Copy function
Remote Mirror and Copy functions, previously known as Peer-to-Peer Remote Copy or
PPRC, which include:
– Metro Mirror, previously known as Synchronous PPRC
– Global Copy, previously known as PPRC Extended Distance
– Global Mirror, previously known as Asynchronous PPRC
– 3-site Metro/Global Mirror with Incremental Resync
z/OS Global Mirror, previously known as Extended Remote Copy (XRC)
z/OS Metro/Global Mirror
You can manage the Copy Services functions through a command-line interface (DS CLI)
and a new Web-based interface (DS Storage Manager). You also can manage the Copy
Services functions through the open application programming interface (DS Open API).
When you manage the Copy Services through these interfaces, these interfaces invoke Copy
Services functions via the Ethernet network. In zSeries environments, you can invoke the
Copy Service functions by TSO commands, ICKDSF, the DFSMSdss™ utility, and so on.
When you set up a FlashCopy operation, a relationship is established between the source
and target volumes, and a bitmap of the source volume is created. Once this relationship and
bitmap are created, the target volume can be accessed as though all the data had been
physically copied. While a relationship between the source and target volume exists,
optionally, a background process copies the tracks from the source to the target volume.
After creating the bitmap, a background process begins to copy the real data from the source
to the target volumes. If you access the source or the target volumes during the background
copy, FlashCopy manages these I/O requests, and facilitates both reading from and writing to
both the source and target copies. When all the data has been copied to the target, the
FlashCopy relationship is ended.
1.1.2 Remote Mirror and Copy RMC (formerly Peer-to-Peer Remote Copy)
The new Remote Mirror and Copy license authorization enables several Copy Services
functions dependent on the configuration and settings used.
The Remote Mirror and Copy feature (RMC) (formerly called Peer-to-Peer Remote Copy, or
PPRC) is a flexible data mirroring technology that allows replication between volumes on two
or more disk storage systems. You can also use this feature for data backup and disaster
recovery. Remote Mirror and Copy is an optional function. To use it, you must purchase the
Remote Mirror and Copy 2244 function authorization model, which is 2244 Model RMC.
Attention: For the DS8000 Turbo models 931, 932, and 9B2, the Metro Mirror and Global
Mirror features are licensed separetely.
DS8000 Storage Units can participate in Remote Mirror and Copy solutions with the ESS
Model 750, ESS Model 800, and DS6000 Storage Units. To establish an RMC (formerly
PPRC) relationship between the DS8000 and the ESS, the ESS needs to have licensed
internal code (LIC) Version 2.4.3.15 or later.
The Remote Mirror and Copy feature can operate in the following modes.
Global Copy does not keep the sequence of write operations. Therefore, the copy is normally
fuzzy, but you can make a consistent copy through synchronization (called a go-to-sync
operation). After the synchronization, you can issue FlashCopy at the secondary site to make
the backup copy with data consistency. After the establishment of the FlashCopy, you can
change the PPRC mode back to the non-synchronous mode.
Chapter 1. Introduction 5
Note: When you change PPRC mode from synchronous to non-synchronous mode, you
change the PPRC mode from synchronous to suspend mode at first, and then you change
PPRC mode from suspend to non-synchronous mode.
If you want make a consistent copy with FlashCopy, you must purchase a Point-in-Time Copy
function authorization (2244 Model PTC) for the secondary Storage Unit.
Global Mirror operations provide the benefit of supporting operations over virtually unlimited
distances between the local and remote sites, restricted only by the capabilities of the
network and the channel extension technology. It can also provide a consistent and
restartable copy of the data at the remote site, created with minimal impact to applications at
the local site.
The ability to maintain an efficient synchronization of the local and remote sites with support
for failover and failback modes helps to reduce the time that is required to switch back to the
local site after a planned or unplanned outage.
This provides a very resilient and flexible solution to recover in various disaster situations.
Customers also benefit from a synchronous replication of the data to a close location acting
as the intermediate site. It also enables the possibility to copy the data across almost
unlimited distance, whereby data consistency can be provided in any time in each location.
With Incremental Resync it is possible to change the copy target destination of a copy relation
without requiring a full copy of the data. This functionality can be used, for example, when an
intermediate site fails due to a disaster. In this case a Global Mirror will be established from
the local to the remote site, which bypasses the intermediate site. When the intermediate site
becomes available again the Incremental Resync is used to bring it back into the
Metro/Global Mirror setup.
The 3-site Metro/Global Mirror is an optional chargeable feature available on all models
(92x/9Ax and 93x/9Bx). It requires Licensed Machine Code (LMC) level 5.2.200.x (bundle
version 6.2.200.x) or later.
You can perform Copy Services operations both within or between Storage Complexes. An
ESS 800 Copy Services Domain (at the correct firmware level) can, for management
purposes, appear to a DS6000 or DS8000 as an independent Storage Complex.
Now within a Storage Complex you will find management consoles, Storage Units, and in the
case of the DS8000, Storage Facility Images. First we define each of these constructs.
In either case, the end-user interacts with the management console using either the Web
browser based DS GUI or the Command-Line Interface (DS CLI). It is possible using the DS
GUI to manage Copy Services operations on multiple Storage Complexes.
Example 2-1 Difference between a DS6000 Storage Unit and DS6000 SFI
dscli> lssu
Date/Time: 13 November 2005 19:21:17 IBM DSCLI Version: 5.1.0.204
Name ID Model WWNN pw state
=========================================================
13-00247 IBM.1750-1300247 511 500507630EFFFE16 On
dscli> lssi
Date/Time: 13 November 2005 19:21:44 IBM DSCLI Version: 5.1.0.204
Name ID Storage Unit Model WWNN State ESSNet
============================================================================
- IBM.1750-1300247 IBM.1750-1300247 511 500507630EFFFE16 Online Enabled
DS8000 SFI
For the DS8000, the SFI concept is necessary because a DS8000 can be ordered as a model
9A2 or 9B2. In these models all resources in the DS8000 are divided between two separate
machine images. This means that to all attached hosts, a single DS8000 Storage Unit
effectively contains two DS8000s. It is like having two separate machines in one physical box.
This means that when using the DS GUI or the DS CLI, you have to be very clear whether
you are working with the SFI or with the Storage Unit. For most operations you always work
with an SFI. You can tell the difference very easily. The serial number of the Storage Unit
always ends in zero. The serial number of the first SFI always ends in 1. If you have ordered
a model 2107-9A2 or 2107-9B2 then you will also have a second SFI. The serial number of
the second SFI always ends in 2.
Example 2-2 Difference between a DS8000 Storage Unit and DS8000 SFI - Model 922
dscli> lssu
Date/Time: 13 November 2005 19:21:01 IBM DSCLI Version: 5.1.0.204
Name ID Model WWNN pw state
=============================================================
2107-7520780 IBM.2107-7520780 922 5005076303FFF9A5 On
dscli> lssi
Date/Time: 13 November 2005 19:21:21 IBM DSCLI Version: 5.1.0.204
Name ID Storage Unit Model WWNN State ESSNet
=================================================================================
ATS_3_EXP IBM.2107-7520781 IBM.2107-7520780 922 5005076303FFC1A5 Online Enabled
Example 2-3 Difference between a DS8000 Storage Unit and DS8000 SFI - Model 9A2
dscli> lssu
Date/Time: 13 November 2005 19:25:25 IBM DSCLI Version: 5.1.0.204
Name ID Model WWNN pw state
================================================================
2107_75ABTV1/V2 IBM.2107-75ABTV0 9A2 5005076303FFFE63 On
dscli> lssi
Date/Time: 13 November 2005 19:25:34 IBM DSCLI Version: 5.1.0.204
Name ID Storage Unit Model WWNN State ESSNet
============================================================================
- IBM.2107-75ABTV1 IBM.2107-75ABTV0 9A2 5005076303FFC663 Online Enabled
- IBM.2107-75ABTV2 IBM.2107-75ABTV0 9A2 5005076303FFCE63 Online Enabled
DS8000 DS8000
DS HMC DS HMC
Ethernet Link
Storage Storage
Complex 1 Complex 2
Note: Only machines of the same type, like 2107, can be in the same Storage Complex.
However, Storage Complexes with different machine types can be joined together. For
example, a DS8000 Storage Complex and a DS6000 Storage Complex can be
inter-connected.
DS HMC DS SMC
Network Interface Network Interface
Server Server
SFI
Network Network Network Network
Copy
Interface Interface Interface Interface
Services
Node Node Node Node
Server
Microcode Microcode
Microcode
SAN A SAN B
ESS Copy
Services
Domain
(Complex 3)
ESS
ESS
800
800
Storage Storage
DS6000
DS6000
DS8000
DS8000
Complex Complex
1 2
HMC1 SMC1
LAN
Client 1 Client 2
The DS GUI is accessed via a Web browser. The DS GUI is not able to save tasks as we do
on the ESS. Thus, you cannot use the DS GUI to initiate a series of saved tasks. To do this we
need to use the DS CLI. If you wish to use a single GUI to manage multiple Storage
Complexes, then you need to define in the GUI all of the Storage Complexes. This will allow
you to manage FlashCopy or Remote Mirror and Copy on, or between, every Storage Unit in
every defined and accessible Storage Complex.
When you look at the structure in Figure 2-2 on page 11, you can see that you need a
working DS SMC or DS HMC in every Storage Complex, to communicate with that complex’s
DS6000 or DS8000. For an inter-Storage Complex Remote Mirror and Copy, the DS GUI will
establish sessions to the source and target Storage Complexes to show all paths and LUNS.
If no DS HMC or DS SMC is available at the remote Storage Complex, you cannot select this
Storage Complex as a target.
It is possible to have two SMCs or two HMCs in a Storage Complex for the purpose of
redundancy.
The ability to send inband commands does not necessarily mean that the DS CLI or DS GUI
do not have a role to play. To be able to issue a command to a DS6000 or DS8000, a System
z operating system needs to be able to communicate with the relevant Storage Unit. We may
have a remote Storage Unit that is not connected via FICON (or ESCON in the case of a
DS8000) to an active z/OS. This means we would not be able to access a conduit CKD
volume on the remote Storage Unit. Now, provided we have TCP/IP connectivity and an
appropriate management station (such as a Windows host with a Web browser or DS CLI
installed), we could use the DS CLI or the DS GUI to manage the remote Storage Unit. We
could then use inband commands to manage the local Storage Unit. If the remote and local
Storage Units were connected by RMC links, then we can also send some commands inband
via the RMC links, such as commands to establish FlashCopies of the remote RMC target
volumes.
Part 2 Interfaces
In this part we discuss the interfaces available to manage the Copy Services features of the
DS8000. We give an overview of the interfaces, discuss the options available, discuss
configuration considerations, and give usage examples of the interfaces.
or
https://10.0.0.1:8452/DS8000
Clearly, you should change the IP address to match the one used by your HMC.
Logging in to the DS Storage Manager requires that you provide your user name and
password. This function is generally administered through your system administrator and by
your company policies. The preconfigured user ID and password are as follows:
User ID: admin
Password: admin
When you log on to the DS GUI for the first time using the admin user ID, you will be forced to
change the password from admin to something more secure.
Then right-click the HMC desktop and select Net → Browser → DS Storage Manager. This
will start the Opera Web Browser. You will not be able to go to any other Web sites from here.
Note: The ability to use the HMC to access the DS GUI means you do not need a separate
PC to configure or manage the DS8000, unless you wish to use the DS CLI. The DS CLI is
not installed on the HMC so it must be installed on an Ethernet attached server or PC.
You still need to understand the implications of the actions you are performing (for example,
you need to understand that a FlashCopy will overwrite the target LUN or volume).
10.0.0.1
A simple scenario may be that we have two sites. At the production site we have a single
DS8000 managed by its own HMC. Then at a remote site we have a second DS8000
managed by its own HMC. There is Ethernet connectivity between the two HMCs. Because
each HMC manages only the DS8000 at its own site, this means we have two Storage
Complexes. If we wish to use a single GUI to manage Copy Services at either site, then we
have to add these Storage Complexes to each other. Once we have done this, we can use
the DS GUI connected to one HMC to manage Copy Services on either Storage Unit at either
site, plus set up Copy Services relations between the two sites.
Note: To use the DS GUI supplied by the HMC in one Storage Complex to manage the
Storage Units of the other Storage Complex, you have to do this operation once in each
direction. This means you have to add Complex1 to Complex2, and then add Complex2 to
Complex1.
Important: Make sure the user ID you use to log on to the DS8000 DS GUI also exists on
the other DS8000 Storage Complex and that it has the same password. If not, the
operation to add the Storage Complex will fail. You will need to always use this same user
ID for multi-complex management.
6. Enter the IP address of the other Storage Complex’s HMC into the Management console 1
IP address box. If the other Storage Complex has a second HMC, select the Define a
second Management console box and enter the second HMC address into the
Management console 2 IP address box. Now select OK.
7. When the add Storage Complex operation completes, the Storage Complex screen will
now show the DS8000 HMC as an additional Storage Complex. In Figure 3-3 we can see
the successful result. We have added a second Storage Complex that manages one
Storage Unit.
10.0.0.1
10.0.0.100
8. Having added one DS8000 Storage Complex to another, you are now able to use the
DS8000 DS GUI to create paths and Remote Mirror and Copy pairs between either of the
Storage Units. You can also use the DS8000 DS GUI to manage FlashCopy pairs on
either DS8000. This is assuming all the relevant licenses are present.
Note: The steps to add one DS8000 Storage Complex to another DS8000 Storage
Complex cannot be performed using the DS CLI.
If you wish to access the public version of the DS8000 Information Center, it can be located
at:
http://publib.boulder.ibm.com/infocenter/ds8000ic/index.jsp
The command-line interface provides a full-function command set that allows you to check
your Storage Unit configuration and perform specific application functions when necessary.
Before you can use the DS CLI commands, you must ensure the following:
It must have been installed as a Full Management Console installation management type.
Your Storage Unit must be configured (part of the DS Storage Manager post-installation
instructions).
You must activate your license activation codes before you can use the DS CLI
commands associated with the Copy Services functions.
The following list highlights a few of the specific types of functions that you can perform with
the DS Command-Line Interface:
Create user IDs that can be used with the GUI and the DS CLI.
Manage user ID passwords.
Install activation keys for licensed features.
Manage Storage Complexes and Units.
Configure and manage Storage Facility Images.
Create and delete RAID arrays, Ranks, and Extent Pools.
Create and delete logical volumes.
Manage host access to volumes.
Check the current Copy Services configuration that is used by the Storage Unit.
Create, modify, or delete Copy Services configuration settings.
Important: For the most recent information about currently supported operating systems,
refer to the IBM System Storage DS6000 Information Center Web site:
http://publib.boulder.ibm.com/infocenter/ds6000ic/index.jsp
The DS CLI is supplied and installed via a CD that ships with the machine. The installation
does not require a reboot of the open systems host. The DS CLI requires Java™ 1.4.1 or
The installation process can be performed via a shell, such as the bash or korn shell, or the
Windows command prompt, or via a GUI interface. If performed via a shell, it can be
performed silently using a profile file. The installation process also installs software that
allows the DS CLI to be completely de-installed should it no longer be required.
If you need any assistance to install the DS CLI you can refer to the publication IBM System
Storage DS8000 Command-Line Interface User´s Guide, SC26-7916.
Important: When a new user is created, the password that was initially set will be
automatically expired and must be changed when the user first logs on.
Attention: The default profile file created when you install the DS CLI will potentially be
replaced every time you install a new version of the DS CLI. It is a better practice to open
the default profile and then save it as a new file. You can then create multiple profiles and
reference the relevant profile file using the -cfg parameter.
These profile files can be specified using the DS CLI command parameter -cfg
<profile_name>. If the -cfg file is not specified, the user’s default profile is used. If a user’s
profile does not exist, the system default profile is used.
Note: A password file generated using the managepwfile command is located at the
following directory: user_home_directory/dscli/profile/security/security.dat.
When you install the command-line interface software, the default profile is installed in the
profile directory with the software. The file name is dscli.profile, for example c:\Program
Files\IBM\DSCLI\profile\dscli.profile.
The available variables and detailed descriptions and information about how to handle them
can be found in the publication IBM System Storage DS8000 Command-Line Interface User´s
Guide, SC26-7916.
lsavailpprcport Lists available ports that can be defined as Remote Mirror and Copy.
showgmir Displays detailed properties and performance metrics for Global Mirror.
showgmiroos Displays the number of unsynchronized (out of sync) tracks for a Global Mirror session.
mkremoteflash Initiates a remote copy through a Remote Mirror and Copy relationship.
mkpprcpath Establishes or replaces a Remote Mirror and Copy path over a Fibre Channel.
mkesconpprcpath Creates a Remote Mirror and Copy path over an ESCON connection.
commitflash Commits data to a target volume to form a consistency between the source and target.
revertflash Overwrites new data with data saved at the last consistency formation.
setflashrevertible Modifies a remote FlashCopy pair that is part of a Global Mirror to revertible.
revertremoteflash Overwrites new data with data saved at the last consistency formation (inband).
setremoteflashrevertible Modifies a remote FlashCopy pair that is part of a Global Mirror to revertible (inband).
failoverpprc Changes a secondary device into a primary suspended and keeps the primary in current.
failbackpprc Usually used after a failoverpprc to reverse the direction of the synchronization.
pausepprc Pauses an existing Remote Mirror and Copy volume pair relationship.
resumepprc Resumes a Remote Mirror and Copy relationship for a volume pair.
Important: The Remote Mirror and Copy commands are asynchronous. This means that a
command is issued to the DS CLI server and if it is accepted successfully, you receive a
successful completion code; however, background activity may still be occurring. For
example, a Metro Mirror pair will take some time to establish, as the tracks need to be
copied across from the primary to secondary device. In this example you should check the
state of the volumes using the lspprc command until the pairs have reached the Duplex
state.
You must supply the login information and the command that you want to process at the
same time. Use the following example to use the single-shot mode:
1. Enter:
dscli -hmc1 <hostname or ip address> -user <adm user> -passwd <pwd> <command>
2. Wait for the command to process and display the end results.
Note: When typing the command, you can use the host name or the IP address of the DS
HMC.
You can issue the DS CLI script from the command prompt at the same time that you provide
your login information:
1. Enter:
dscli -hmc1 <hostname ip address> -user <user name> -passwd <pwd> -script <full path of
script file>
Important : The DS CLI script can contain only DS CLI commands. Use of shell
commands results in process failure. You can add comments to the scripts prefixed by the
number sign (#).
When typing the command, you can use the host name or the IP address of the DS HMC.
Note: When typing the command, you can use the host name or the IP address of the DS
HMC.
The return codes used by the DS CLI are shown in Table 4-6.
In Example 4-5, the -l parameter is used to get a list of all parameters that can be used with
the mkflash command.
Man pages
A man page is available for every DS CLI command. Man pages are most commonly seen in
UNIX-based operating systems to give information about command capabilities. This
information can be displayed by issuing the relevant command followed by -h, -help, or -?, for
example:
dscli> mkflash -help
or
dscli> help mkflash
Example 4-6 shows some of the common commands used on a DS6000. (The same
commands are possible on the DS8000.)
Part 3 FlashCopy
This part of the book describes the IBM System Storage FlashCopy when used in open
systems environments with the DS8000. We discuss the FlashCopy features and describe
the options for its setup. We also show which management interfaces can be used as well as
the important aspects to be considered when establishing FlashCopy relationships.
This part ends with examples showing how FlashCopy is used in the daily business to
support the various environments where immediate copies of production data are a
requirement.
This chapter explains the basic characteristics of FlashCopy when used in an open systems
environment with the DS8000. The following topics are discussed:
FlashCopy operational areas
FlashCopy basic concepts
FlashCopy in combination with other Copy Services
FlashCopy in a storage LPAR environment
Production
Production Backup
Application
System System
Help desk
System or
Operation System
Reverse Operation
FlashCopy Data
Integration
Backup
System
System
Business Production System
Integration data Operation
With the capability to reverse a FlashCopy, a previously created FlashCopy can be used
within seconds to bring production back to the level of data it had at the time when the
FlashCopy was taken.
5.2 Terminology
When discussing about Metro Mirror, Global Copy and Global Mirror, you will see that the
following terms are frequently used and interchangeably:
The terms local, production, application, primary, or source, denote the site where the
production applications run while in normal operation. These applications create, modify,
and read the data — the application data. The meaning is extended to the disk subsystem
that holds the data as well as to its components — volumes and LSS.
The terms remote, recovery, backup, secondary, or target, denote the site to where the
data is replicated — the copy of the application data. The meaning is extended to the disk
subsystem that holds the data as well as to its components — volumes and LSS.
When discussing about FlashCopy we use the term source to refer to the original data that is
created by the application. And we use the term target to refer to the point-in-time backup
copy.
Also the terms LUN and volume are used interchangeably in our discussions.
As result of the FlashCopy either all physical blocks from the source volume are copied (full
copy) or — when using the nocopy option — only those parts that are changing in the source
data since the FlashCopy has been established. Currently, the target volume needs to be the
same size or bigger than the source volume whenever FlashCopy is used to flash a whole
volume.
Typically, large applications such as databases have their data spread across several
volumes and their volumes should all be FlashCopied at exactly the same point-in-time.
FlashCopy offers consistency groups, which allows multiple volumes to be FlashCopied at
exactly the same instance.
time
t0
bitmap
1 1 1 1 1 1
data data
t0 t0 t0 t0 t0 t0
Once the relationship has been established, it is possible to perform read and write I/Os
on both the source and the target. Assuming that the target is used for reads only while
production is ongoing, things will look as illustrated in Figure 5-3.
time
t0 tx ty tz
read 1 read 2
1 0 1 0 1 1
data data
t0 tx t0 tz t0 t0 t0 t0
Figure 5-3 Reads from source and target volumes and writes to source volume
time
t0 ta tb
write a write b
bitmap
0 0 1 0 1 1
data data
t0 tx t0 ty t0 t0 ta t0 tb
If not explicitly defined as persistent, the FlashCopy relationship ends as soon as all data is
copied.
Background copy
time
t0 tx ty
bitmap
0 0 0 0 0 0
data data
t0 tx t0 ty t0 t0 t0 t0 t0 t0 t0 t0
Background copy will copy all time-zero data from source volume to target volume
Figure 5-5 Target volume after full volume FlashCopy relationship finished
If there were writes to the target then the picture we get is similar to the one in Figure 5-6.
Background copy
time
t0 tx ty ta tb
bitmap
0 0 0 0 0 0
data data
t0 tx t0 ty t0 t0 ta t0 t0 tb t0 t0
Background copy will copy all time-zero data from source volume to target volume.
If the persistent FlashCopy option was specified, the FlashCopy relationship must be
withdrawn explicitly.
Note: The sample scenarios discussed in the present section, only apply to full-volume
FlashCopy. Data set FlashCopy is not supported in the open systems environment.
FlashCopy
source target
Metro
Mirror
primary
source target
FlashCopy
Figure 5-7 FlashCopy and Metro Mirror
As illustrated in Figure 5-7, at the primary Metro Mirror site, the following combinations are
supported:
A FlashCopy source volume can become a Metro Mirror primary volume and vice versa.
The order of creation is optional.
A FlashCopy target volume can become a Metro Mirror primary volume and vice versa. If
you wish to use a FlashCopy target volume as a Metro Mirror primary, be aware of the
following considerations:
– The recommended order is to first establish the Metro Mirror, and then create a
FlashCopy to that Metro Mirror primary using the -tgtpprc parameter.
The Metro Mirror secondary will not be in a fully consistent state until the Metro Mirror
enters the full duplex state.
On the secondary site of the Metro Mirror, a FlashCopy source volume can be the Metro
Mirror secondary volume, and vice versa. There are no restrictions on which relationship
should be defined first.
FlashCopy
source target
Global
primary
Copy
source target
FlashCopy
Figure 5-8 FlashCopy and Global Copy
As illustrated in Figure 5-8, at the primary Global Copy site the following combinations are
possible:
A FlashCopy source volume can become a Global Copy primary volume and vice versa.
The order of creation is optional.
A FlashCopy target volume can become a Global Copy primary volume and vice versa. If
you want to use a FlashCopy target volume as a Global Copy primary, be aware of the
following considerations:
– The recommended order is to first establish the Global Copy, and then create a
FlashCopy to that Global Copy primary using the -tgtpprc parameter.
The Global Copy secondary is will not be in a fully consistent state until the Global
Copy enters the full duplex state.
Execute the mkpprc -type mmir command to force the Global Copy to enter the full
duplex state.
– If you create the FlashCopy first and then do a Global Copy of the FlashCopy target,
you must monitor the progress of the FlashCopy background copy.
On the secondary site of the Global Copy a FlashCopy source volume can be based on the
secondary volume for the Global Copy.
FlashCopy
source target
secondary Global
Mirror
primary for
Global open
Mirror for open
Figure 5-9 FlashCopy and Global Mirror
A FlashCopy source volume can become a Global Mirror primary volume and vice versa. The
relationships can be established in any sequence. A FlashCopy target volume cannot
become a Global Mirror primary volume.
On the Global Mirror secondary site, the Global Mirror target volume cannot be used as
FlashCopy source or FlashCopy target unless the Global Mirror pair is first suspended.
Processor Processor
complex 0 complex 1
Storage
LPAR01 Image 1 LPAR11
Storage
LPAR02 Image 2 LPAR12
Note in the figure that two processor complex LPARs, one from each processor complex,
make one storage LPAR. That is, processor complex LPAR 01 and LPAR 11 make the
DS8000 storage LPAR 1 — storage image 1.
For each source volume, only one FlashCopy relationship can be reversed (the one having
the -record attribute).
A FlashCopy target volume can have only one FlashCopy source volume.
A FlashCopy target volume cannot be a FlashCopy source volume at the same time.
Figure 6-1 illustrates what is possible and what is not with multiple relationship FlashCopy.
FlashCopy
source target
...
maximum is 12
...
FlashCopy FlashCopy
source target
source and
source target target
With the Freeze FlashCopy Consistency Group option, the DS8000 holds off I/O activity to a
volume for a time period by putting the source volume in a queue full state. Therefore, a time
slot can be created during which dependent write updates do not occur, and FlashCopy uses
that time slot to obtain a consistent point-in-time copy of the related volumes. I/O activity
resumes when all FlashCopies are established.
Dependent writes
If the start of one write operation is dependent upon the completion of a previous write, the
writes are dependent. Application examples for dependent writes are databases with their
associated logging files. For instance, the database logging file will be updated after a new
entry has been successfully written to a tablespace. The chronological order of dependent
writes to the FlashCopy source volumes is the basis to provide consistent data at the
FlashCopy target volumes. For a more detailed understanding of dependent writes and how
the DS8000 enables the creation of consistency groups, thus ensuring data integrity on the
target volumes, you can refer to the discussion in 12.4.1, “Data consistency and dependent
writes” on page 117.
Figure 6-2 illustrates this capability. In this figure, the FlashCopy target and the Metro Mirror
(or Global Copy) primary are the same volume. They are displayed as two separate volumes
for ease of understanding.
source target
FlashCopy
primary secondary
Figure 6-2 FlashCopy target is Metro Mirror (or Global Copy) primary
It is possible to either create the FlashCopy relationship first or the Metro Mirror (or Global
Copy) relationship first. However, in general it is better to create the FlashCopy relationship
first to avoid the initial sending of unnecessary data across to the Metro Mirror (or Global
Copy) secondary.
Important: A refresh of the target volume always overwrites any writes previously written
to the target volume.
To perform an incremental FlashCopy, you must first establish the FlashCopy relationship
with the Start Change Recording and Persistent FlashCopy options enabled.
nothing to be
done as data
was already
copied from
no updates in source source to target no updates in target
and is identical
on both sides
Figure 6-3 Updates to the target volume caused by a refresh target FlashCopy
With Incremental FlashCopy, the initial FlashCopy — copy or nocopy — relationship between
a source and target volume is subject to the following (see Figure 6-3):
FlashCopy with nocopy option
If the original FlashCopy was established with the nocopy option, then the bitmap for the
target volume will be reset, and of course, the updates on the target volumes are
overwritten. However, using the incremental option will automatically convert this
relationship to a copy relationship, and the background copy will begin.
FlashCopy with copy option
If the original FlashCopy was established with the copy option (full volume copy), then the
updates that took place on the source volume since the last FlashCopy will be copied to
the target volume. Also, the updates done on the target volume will be overwritten with the
contents of the source volume.
These bitmaps allow subsequent FlashCopies to transmit only those blocks of data for which
updates occurred. Every write operation to the source or target volume will be reflected in
these bitmaps by setting the corresponding bit to 0.
time
t0 tx ty tz ta read 1 read 2 t b tc
write y write a time-zero data write b write c
read write x write z not yet available
in target :
read it from
bitmap source.
bitmap
0 0 1 0 0 1 0 0 1 0 0 0
source target
data data
t0 tx t0 tz t0 t0 ta t0 t tb t0 t
time
t0 tx ty tz ta tb t c t 0'
bitmap bitmap
0 0 1 0 0 1 0 0 1 0 0 0
source target
data data
t0 tx t0 tz t0 t0 t0 tx t0 tz t0 t0
needs to be copied as a
write occured on the target
update to the source
needs to be copied
update to the source
needs to be copied
needs to be copied as a
write occured on the target
bitmap
1 1 1 1 1 1 1 1 1 1 1 1
source target
data data
After the refresh (which takes place only on the bitmap level) the new FlashCopy based on
time-0’ is active. The copy of the time-0’ data to the target is done in the background.
Tip: You can do the incremental copy at any time. You do not have to wait for the previous
background copy to complete.
The FlashCopy source volume at the remote site must be the secondary volume of the Metro
Mirror or Global Copy pair.
primary secondary
source target
FlashCopy
Figure 6-6 illustrates this capability. In this figure, the Metro Mirror (or Global Copy) secondary
and the FlashCopy source are the same volume. They are displayed as two separate volumes
for ease of understanding.
nothing to be done
as data was already
copied from source
no updates in source to target and is no updates in target
identical on both
sides
The source and target bitmaps (illustrated in Figure 6-4 on page 51) are exchanged and then
handled as described with the Incremental FlashCopy option.
Remote FlashCopy
Persistent Flashcopy
Reverse restore,
fast reverse restore
Note: For a detailed explanation of the features involved and the considerations you must
have when ordering FlashCopy we recommend that you refer to the announcement letters:
IBM System Storage DS8000 — Functions Authorization (IBM 2244)
IBM System Storage DS8000 Series (IBM 2107)
All DS8000 series licensed functions are enabled, authorized, managed, activated, and
enforced based upon the physical capacity contained in a 2107 system.
Enablement
Licensed functions are enabled for a DS8000 through the selection of the appropriate
licensed function indicator feature number on the 2107 base unit. Particularly for FlashCopy,
the 2107 feature that must be ordered is the Licensed Function Indicator - Point-in-time copy
(PTC) #0720.
Authorization
To use FlashCopy you must acquire the corresponding DS8000 Series Function
Authorization (IBM 2244-PTC) with the adequate feature number (#72xx) to establish the
extent of IBM authorization: license size in terms of physical capacity. The DS8000 Series
Function Authorizations are for billing purposes only and establish the extent of IBM
authorization for use of a particular licensed function on the IBM System Storage DS8000
series (IBM 2107).
The feature activation codes are made available by IBM and are obtained using the Disk
Storage Feature Activation (DSFA) application at:
http://www.ibm.com/storage/dsfa
7.2.1 Management
The Disk Storage Feature Activation (DSFA) application can be used to:
Select the license scope.
Assign the license value.
Deactivate a licensed function.
The previous activities are referred to as the management activities that you can perform with
the DSFA application and in relation to the DS8000 licensed functions.
In particular, the license scope of FlashCopy is multiple: it can either be FB, CKD, or ALL. For
this reason when ordering FlashCopy you must plan ahead what your point-in-time copy
requirements will be. Example 7-1 illustrates a sample case.
If a licensed function (as is the case with FlashCopy) has multiple license scope options, you
will be required to select a license scope during the initial retrieval of the feature activation
code.
Note: The license scope is not selected while ordering function authorization feature
numbers. The feature numbers only establish the extent of IBM authorization (in terms of
physical capacity) regardless of the storage type.
You can also change the license scope using the DSFA application after a licensed function
has been activated. A new feature activation code will be generated, and when you install it
into the machine, the function will be activated and enforced using the newly selected license
scope. Only an increase in the license scope (changing FB or CKD to ALL) is a nondisruptive
activity. A lateral change (changing FB to CKD or changing CKD to FB) or a reduction of the
license scope (changing ALL to FB or CKD) is a disruptive activity and requires a machine
IML.
On the Model 9B2, the license scope does not need to be the same for each IBM System
Storage DS8300 LPAR. For example, with the FlashCopy license function you can select a
license scope of FB for Storage System LPAR 1 and a license scope of ALL for Storage
System LPAR 2. You can also change the license scope on a Storage System LPAR basis
and, if the change is disruptive, you will only need to reboot the affected Storage System
LPAR.
On the DS8000 Models 931 and 932, license value assignment is an optional activity you can
perform using the DSFA application. On these models, the feature activation codes will
always be generated using the total license value for each licensed function, unless you
assign a license value of zero (0.0 TB).
On the Model 9B2, license value assignment is a required activity that you perform using the
DSFA application. On this model, the license value does not need to be the same for each
Storage System LPAR. For example, if you have a 15 TB total authorization level for the PTC
licensed function, you can assign a license value of 9.1 TB to Storage System LPAR 1 and a
license value of 5.9 TB to Storage System LPAR 2.
You can also change the license value on a Storage System LPAR basis. If the change is a
reduction in the license value, you may need to reboot the affected Storage System LPAR.
Deactivation
Assigning a license value of zero (0.0 TB) provides the ability to deactivate (disable) a
licensed function. A new feature activation code will be generated, and when you install it into
the machine, the function will be deactivated during the next IML of the machine.
You can also reactivate a deactivated licensed function by changing the license value to a
non-zero value. A new feature activation code will be generated, and when you install it into
the machine, the function will be activated. Reactivation is a nondisruptive activity.
When you deactivate a licensed function using the DSFA application, the installed function
authorization feature numbers remain on the machine record. This means that you do not
need to reacquire your function authorization feature numbers if you want to reactivate the
licensed function at a future date.
7.2.2 Activation
Activation refers to the client retrieval and installation of the feature activation code into the
2107 system.
The feature activation code is made available by IBM and obtained using the DSFA
application at:
http://www.ibm.com/storage/dsfa
Note: To access DSFA, you will be required to enter information about your 2107 machine.
You can find this information about the Storage Unit General Properties page in the IBM
System Storage DS8000 Storage Manager application.
The following information is needed for the download of the keyfile and should be prepared
during planning:
Model
The model of the DS8000 can be taken from the order.
Serial number of the DS8000
The serial number of the DS8000 can be taken from the front of the base frame (lower
right corner). Especially if several machines have been delivered, this is the only way to
obtain the serial number of a machine located in a specific point in the computer center.
This chapter gives an overview of the DS CLI and the DS SM for FlashCopy management.
DS CLI and DS SM are also covered in Part 2, “Interfaces” on page 15.
Also, the DS Open Application Programming Interface (DS Open API), which is a set of
application programming interfaces that are available to be integrated in programs, can be
used for FlashCopy management and control. The DS Open API is not covered in the present
book. For information about the DS Open API, refer to the publication IBM System Storage
DS Open Application Programming Interface Reference, GC35-0516.
The DS CLI can be used to execute both local FlashCopy and remote FlashCopy commands.
The DS SM user interface and the DS Open API do not support remote FlashCopy.
Local FlashCopy refers to the situation where the FlashCopy is local as opposed to a remote
FlashCopy. For a discussion of the characteristics of a remote FlashCopy refer to 6.5
“Remote FlashCopy” on page 52.
Create a FlashCopy
Modify a FlashCopy pair that is part of a Global Mirror setflashrevertible restorable, Record
relationship to revertible Change Wizard
Run new background copy for persistent FlashCopy rmflash -cp background copy
Terminate FlashCopy
Create a FlashCopy
Terminate FlashCopy
Note: Remote FlashCopy is not supported with the DS SM or DS Open API interfaces.
DS CLI Commands
mkflash lsflash setflash commit resync reverse revert unfreeze rmflash
Parameters revertible flash flash flash flash flash
Source
freeze x x
Target
tgtpprc x x x
tgtoffline x x x x
tgtinhibit x x x
tgtonly x
Flashcopy pair
dev x x x x x x x x x
record x x x x
persist x x x x
nocp x
seqnum x x x x x x x x
source:target x x x x
fast x
source x x
cp x x
source LSS x
l x
s x
activecp x
revertible x
Command
wait x
quiet x
Figure 8-1 summarizes the parameters and the corresponding DS CLI commands. When
FlashCopy receives these parameters, the following actions result:
freeze: Consistency Group FlashCopy.
With the DS CLI, it is possible to establish a Consistency Group by using the -freeze
parameter and identifying all FlashCopy pairs and target volumes belonging to it.
tgtpprc: establish target on existing Metro Mirror source.
When this option is selected, the target volume can be or become a source volume for a
Metro Mirror or Global Copy relationship.
tgtoffline: permit FlashCopy to occur if target volume is not online for host access.
When setting up the relationship, a verification is done to see whether the target volume is
online or offline. This parameter only applies to CKD volumes. Table 8-3 summarizes the
actions that result depending on the target volume status.
The background copy takes place immediately and the relationship remains as a
persistent relationship. Using other DS CLI commands it could be reversed and
re-synchronized.
Example 3: 0003 → 0103
The FlashCopy between volume 0003 and volume 0103 is established with the following
FlashCopy properties enabled: Persistent and BackgroundCopy. The background copy
takes place immediately. Once the background copy has finished the FlashCopy
relationship will remain, because of the persistent flag.
Example 4: 0004 → 0104
The FlashCopy between volume 0004 and volume 0104 is established with the -nocp
parameter. This means, no full volume background copy will be done — only the data
Example 8-2 shows a script with several lsflash commands and the output of the script (this
script is logically based on the example for mkflash).
#--- Example 2
lsflash -dev IBM.2107-7506571 0001-0005
Date/Time: July 11, 2005 6:40:29 PM CEST IBM DSCLI Version: 5.0.3.134 DS: IBM.2107-7506571
ID SrcLSS SequenceNum Timeout ActiveCopy Recording Persistent Revertible SourceWriteEnabled TargetWriteEnabled BackgroundCopy
====================================================================================================================================
0003:0103 00 3 300 Disabled Enabled Enabled Disabled Disabled Disabled Enabled
0004:0104 00 4 300 Disabled Disabled Enabled Disabled Disabled Disabled Enabled
0005:0105 00 5 300 Disabled Disabled Disabled Disabled Disabled Disabled Disabled
#--- Example 3
lsflash -dev IBM.2107-7506571 -l 0001-0005
Date/Time: July 11, 2005 6:40:52 PM CEST IBM DSCLI Version: 5.0.3.134 DS: IBM.2107-7506571
ID SrcLSS SequenceNum Timeout ActiveCopy Recording Persistent Revertible SourceWriteEnabled TargetWriteEnabled
BackgroundCopy OutOfSyncTracks DateCreated DateSynced
=======================================================================================================================
0003:0103 00 3 300 Disabled Enabled Enabled Disabled Disabled Disabled
Enabled 0 Mon Jul 11 19:30:06 CEST 2005 Mon Jul 11 19:30:06 CEST 2005
0004:0104 00 4 300 Disabled Disabled Enabled Disabled Disabled Disabled
Enabled 0 Mon Jul 11 19:30:10 CEST 2005 Mon Jul 11 19:30:10 CEST 2005
0005:0105 00 5 300 Disabled Disabled Disabled Disabled Disabled Disabled
Disabled 50085 Mon Jul 11 19:30:13 CEST 2005 Mon Jul 11 19:30:13 CEST 2005
#--- Example 4
lsflash -dev IBM.2107-7506571 -s 0001-0005
Date/Time: July 11, 2005 6:40:59 PM CEST IBM DSCLI Version: 5.0.3.134 DS: IBM.2107-7506571
ID
=========
0003:0103
0004:0104
0005:0105
#--- Example 5
lsflash -dev IBM.2107-7506571 -activecp 0001-0004
Date/Time: July 11, 2005 6:41:02 PM CEST IBM DSCLI Version: 5.0.3.134 DS: IBM.2107-7506571
#--- Example 6
lsflash -dev IBM.2107-7506571 -record 0001-0004
Date/Time: July 11, 2005 6:41:08 PM CEST IBM DSCLI Version: 5.0.3.134 DS: IBM.2107-7506571
ID SrcLSS SequenceNum Timeout ActiveCopy Recording Persistent Revertible SourceWriteEnabled TargetWriteEnabled BackgroundCopy
====================================================================================================================================
0003:0103 00 3 300 Disabled Enabled Enabled Disabled Disabled Disabled Enabled
#--- Example 8
lsflash -dev IBM.2107-7506571 -revertible 0001-0004
Date/Time: July 11, 2005 6:41:22 PM CEST IBM DSCLI Version: 5.0.3.134 DS: IBM.2107-7506571
#--- Example 9
lsflash -dev IBM.2107-7506571 -cp 0001-0004
Date/Time: July 11, 2005 6:41:32 PM CEST IBM DSCLI Version: 5.0.3.134 DS: IBM.2107-7506571
ID SrcLSS SequenceNum Timeout ActiveCopy Recording Persistent Revertible SourceWriteEnabled TargetWriteEnabled BackgroundCopy
====================================================================================================================================
0003:0103 00 3 300 Disabled Enabled Enabled Disabled Disabled Disabled Enabled
0004:0104 00 4 300 Disabled Disabled Enabled Disabled Disabled Disabled Enabled
The following explanations apply to the cases presented in Example 8-2 on page 70:
Example 1: list FlashCopy information for a specific volume.
In this example, the lsflash command shows the FlashCopy relationship information for
volume 0004, showing the status (enabled/disabled) of the FlashCopy properties.
Example 2: list existing FlashCopy relationships information within a range of volumes.
In this example, the lsflash command shows the FlashCopy relationships information for
the range of volumes 0001 to 0005, showing the properties status (enabled/disabled).
Example 3: list existing FlashCopy relationships with full information.
Using the -l parameter with the lsflash command displays the default output plus
information about the following properties: OutOfSyncTracks, DateCreated, and
DateSynced.
Example 4: list ids of existing FlashCopy pairs within a volume range.
Using the -s parameter displays only the FlashCopy source and target volume IDs for the
specified range of volumes.
Example 5: list FlashCopy relationships with active background copy running.
Using the parameter -activecp will display only those FlashCopy relationships within the
selected range of volumes for which a background copy is actively running. The output
format is the default output. In our example there were no active background copies.
Example 6: list existing FlashCopy relationships with -record enabled.
Using the -record parameter will display only those FlashCopy relationships within the
selected range of volumes that were established with the -record parameter.
Example 7: list existing FlashCopy relationships with the Persistent attribute enabled.
When using the parameter -persist, only those FlashCopy relationships within the range of
selected volumes for which the Persistent option is enabled will be displayed.
Example 8: list existing FlashCopy relationships which are revertible.
When using the parameter -revertible, only those FlashCopy relationships within the range
of selected volumes for which the option Revertible is enabled will be displayed. There
were no revertible relationships in our example.
Example 9: list existing FlashCopy relationships for which BackgroundCopy is enabled.
The FlashCopy properties Recording and Persistent must be enabled to set a FlashCopy
relationship to revertible using this command. This command needs to be executed prior to
running a commitflash or revertflash.
Example 8-3 illustrates two situations when using the setflashrevertible command.
#--- Example 2
setflashrevertible -dev IBM.2107-7506571 0003:0103
Date/Time: July 11, 2005 9:53:07 PM CEST IBM DSCLI Version: 5.0.3.134 DS: IBM.2107-7506571
CMUN03027E setflashrevertible: 0003:0103: FlashCopy operation failure: action prohibited by current FlashCopy state
Before the FlashCopy relationship can be committed, it needs to be made revertible. Typically,
this is done automatically by an application such as Global Mirror. However, it can also be set
manually.
#--- Example 2
commitflash -dev IBM.2107-7506571 0003:0103
CMUN03027E commitflash: 0003:0003: FlashCopy operation failure: action prohibited by current FlashCopy state
Tip: You do not have to wait for the background copy to complete before you do the
FlashCopy re-synchronization. The resyncflash command can be used at any time.
#--- Example 2
mkflash -dev IBM.2107-7506571 -nocp -seqnum 03 0004:0104
Date/Time: July 11, 2005 11:01:41 PM CEST IBM DSCLI Version: 5.0.3.134 DS: IBM.2107-7506571
CMUC00137I mkflash: FlashCopy pair 0004:0104 successfully created.
lsflash -dev IBM.2107-7506571 0004
Date/Time: July 11, 2005 11:01:45 PM CEST IBM DSCLI Version: 5.0.3.134 DS: IBM.2107-7506571
ID SrcLSS SequenceNum Timeout ActiveCopy Recording Persistent Revertible SourceWriteEnabled TargetWriteEnabled BackgroundCopy
====================================================================================================================================
0004:0104 00 3 300 Disabled Disabled Disabled Disabled Disabled Disabled Disabled
resyncflash -dev IBM.2107-7506571 -record -persist -seqnum 14 0004:0104
Date/Time: July 11, 2005 11:01:58 PM CEST IBM DSCLI Version: 5.0.3.134 DS: IBM.2107-7506571
CMUN03027E resyncflash: 0004:0104: FlashCopy operation failure: action prohibited by current FlashCopy state
#--- Example 1
lsflash -dev IBM.2107-7506571 0000-0005
Date/Time: July 11, 2005 11:28:33 PM CEST IBM DSCLI Version: 5.0.3.134 DS: IBM.2107-7506571
ID SrcLSS SequenceNum Timeout ActiveCopy Recording Persistent Revertible SourceWriteEnabled TargetWriteEnabled BackgroundCopy
====================================================================================================================================
0001:0101 00 1 300 Disabled Enabled Enabled Disabled Disabled Disabled Enabled
0002:0102 00 2 300 Disabled Enabled Enabled Disabled Disabled Disabled Enabled
0003:0103 00 3 300 Disabled Enabled Enabled Disabled Disabled Disabled Enabled
0005:0105 00 1 300 Disabled Enabled Enabled Disabled Disabled Disabled Enabled
reverseflash -dev IBM.2107-7506571 -record -persist 0001:0101 0005:0105
Date/Time: July 11, 2005 11:33:21 PM CEST IBM DSCLI Version: 5.0.3.134 DS: IBM.2107-7506571
CMUC00169I reverseflash: FlashCopy volume pair 0001:0101 successfully reversed.
CMUC00169I reverseflash: FlashCopy volume pair 0005:0105 successfully reversed.
reverseflash -dev IBM.2107-7506571 -record 0002:0102
Date/Time: July 11, 2005 11:33:27 PM CEST IBM DSCLI Version: 5.0.3.134 DS: IBM.2107-7506571
CMUC00169I reverseflash: FlashCopy volume pair 0002:0102 successfully reversed.
reverseflash -dev IBM.2107-7506571 0003:0103
Date/Time: July 11, 2005 11:33:33 PM CEST IBM DSCLI Version: 5.0.3.134 DS: IBM.2107-7506571
CMUC00169I reverseflash: FlashCopy volume pair 0003:0103 successfully reversed.
lsflash -dev IBM.2107-7506571 0000-0005
Date/Time: July 11, 2005 11:33:37 PM CEST IBM DSCLI Version: 5.0.3.134 DS: IBM.2107-7506571
ID SrcLSS SequenceNum Timeout ActiveCopy Recording Persistent Revertible SourceWriteEnabled TargetWriteEnabled BackgroundCopy
====================================================================================================================================
0101:0001 01 1 300 Disabled Enabled Enabled Disabled Disabled Disabled Enabled
0102:0002 01 2 300 Disabled Enabled Enabled Disabled Disabled Disabled Enabled
0103:0003 01 3 300 Disabled Disabled Disabled Disabled Disabled Disabled Enabled
0105:0005 01 1 300 Disabled Enabled Enabled Disabled Disabled Disabled Enabled
#--- Example 2
reverseflash -dev IBM.2107-7506571 -record 0002:0102
Date/Time: July 11, 2005 11:42:17 PM CEST IBM DSCLI Version: 5.0.3.134 DS: IBM.2107-7506571
CMUC00169I reverseflash: FlashCopy volume pair 0002:0102 successfully reversed.
lsflash -dev IBM.2107-7506571 0002
Date/Time: July 11, 2005 11:42:30 PM CEST IBM DSCLI Version: 5.0.3.134 DS: IBM.2107-7506571
ID SrcLSS SequenceNum Timeout ActiveCopy Recording Persistent Revertible SourceWriteEnabled TargetWriteEnabled BackgroundCopy
====================================================================================================================================
0102:0002 01 2 300 Disabled Enabled Enabled Disabled Disabled Disabled Enabled
#--- Example 3
reverseflash -dev IBM.2107-7506571 -seqnum 12 -record 0102:0002
Date/Time: July 11, 2005 11:46:34 PM CEST IBM DSCLI Version: 5.0.3.134 DS: IBM.2107-7506571
CMUC00169I reverseflash: FlashCopy volume pair 0102:0002 successfully reversed.
lsflash -dev IBM.2107-7506571 0002
Date/Time: July 11, 2005 11:46:39 PM CEST IBM DSCLI Version: 5.0.3.134 DS: IBM.2107-7506571
ID SrcLSS SequenceNum Timeout ActiveCopy Recording Persistent Revertible SourceWriteEnabled TargetWriteEnabled BackgroundCopy
====================================================================================================================================
0002:0102 00 12 300 Disabled Enabled Enabled Disabled Disabled Disabled Enabled
In the example presented in Example 8-8, create new background copy of a persistent
FlashCopy relationship. The existing FlashCopy relationship 0001:0101 is used to create a
new background copy.
In scripts it should always be used with the -quiet parameter to avoid the confirmation prompt.
See Example 8-9.
Table 8-4 shows the similarity between both sets of commands — their actions also similar.
DS CLI Commands
mkremote lsremote setremote commit resync reverse revert rmflash
flash flash flash remote remote remote remote
Parameters revertible flash flash flash flash
Source
freeze x x
Target
tgtpprc x x x
tgtoffline x x x x
tgtinhibit x x x
tgtonly x
Flashcopy pair
dev x x x x x x x x
record x x x x
persist x x x x
nocp x
seqnum x x x x x x x x
source:target x x x x
fast x
source x x
cp x x
source LSS
l x
s x
activecp x
revertible x
Command
conduit x x x x x x x x
quiet x
Most of the parameters for remote FlashCopy commands are similar to those for local
FlashCopy commands. Two major exceptions between local FlashCopy and remote
FlashCopy commands are:
Each remote command has the parameter conduit to identify the link to pass the
command to the secondary DS8000.
The local FlashCopy command unfreezeflash has no remote equivalent.
Figure 8-2 summarizes the parameters and the corresponding DS CLI commands that can be
used when doing remote FlashCopy.
The description of the parameters is similar to the description presented in 8.3.1 “Parameters
used with local FlashCopy commands” on page 67. In regard to the parameter conduit, which
only applies to remote FlashCopy, the following explanation applies: Within a remote mirror
environment, the FlashCopy commands are sent across the mirror paths, thus avoiding the
necessity of having separate network connections to the remote site solely for the
To start using the DS SM, enter in your Web browser the IP address of the DS HMC, as
shown in Example 8-12. The password is set up by the administrator and needs to be
changed during the first use of the DS SM.
After you login to the DS SM, follow this sequence (refer to Figure 8-3):
1. Select Real-time manager.
2. Select Copy services.
3. Select FlashCopy.
4. Use the Storage complex, Storage unit, and Storage image pull-down windows to identify
the DS8000 for which you would like to initiate a FlashCopy. Then use the Resource type
and Specify LSS pull-downs to display the volumes you want to work with on that DS8000.
5. From the Select action pull-down, select Create.
5. Verification
In the verification window (not shown), all information regarding the FlashCopy definition is
displayed. If it is not possible to establish the FlashCopy relationship (if, for example, a
volume was requested to be offline, but is not offline), then an informational message is
displayed on this window.
Using the DS SM, it is possible to define FlashCopies that will be generated immediately. It is
currently not possible to store any FlashCopy tasks using the DS SM.
Table 8-5 Comparison of options and parameters used for FlashCopy in DS CLI and DS SM
Options Parameter Parameter with Comments
with DS CLI DS SM FlashCopy create
command
mkflash
Permit target volume to be online tgtoffline Permit FlashCopy to occur if This option exists only for CKD
in z/OS environment target is online for host access volumes (z/OS)
This will give you a list of all active FlashCopy relationships. By checking the box left of the
FlashCopy of interest, the Select Actions for this FlashCopy relationship are changed based
on the attributes of the FlashCopy relationship and the possible actions to perform on it. Using
the Select Action → Properties, all attributes of the selected FlashCopy are displayed.
Source LSS
Sequence Number to which the FlashCopy belongs. Can also be used for Consistency Groups.
Shows if a Background Copy is currently running. If the DS CLI returns ActiveCopy=disabled this
could be similar to out-of-synch tracks>0 or copy complete on the DS SM.
Shows if recording was selected for both source and target volume to allow for later usage of resync.
Identifies if the FlashCopy relationship can be changed by copying the contents of the target volume
to the source volume
Identifies it the target can be used by another server to write on it in parallel to the FlashCopy taking
place
Then by checking the box at the left of the FlashCopy of your interest, the available Select
Actions for this FlashCopy relationship will be shown. If the existing FlashCopy relationship
can be reversed, it is possible to choose the Select Action and then select Reverse
relationship.
The next window that is presented is shown in Figure 8-10. It prompts for the FlashCopy pairs
for which the background copy should run.
The prompt window asks for more details for the resync request (see Figure 8-12).
Figure 8-12 Prompt window to detail resync request for FlashCopy relationship
The next window is a prompt asking you to confirm the delete request (see Figure 8-14).
Figure 8-14 Prompt window to confirm delete request for FlashCopy relationship
This chapter examines when to use copy versus no copy and where to place the FlashCopy
source and target volumes/LUNs. We also discuss when and how to use incremental
FlashCopy, which should definitely be evaluated for use in most applications.
Note: This chapter is equally valid for System z volumes and Open Systems LUNs. In the
following sections of the present chapter only the terms volume or volumes will be used,
but the text is equally valid if the terms LUN and LUNs were used, unless otherwise noted.
Terminology
Before proceeding with the discussion on FlashCopy best practices, let us review some of the
basic terminology we use in this chapter.
Server: In a DS8000, a server is effectively the software that uses a logical partition
(LPAR) and that has access to a percentage of the memory and processor resources
available on a processor complex. The DS8000 models 931 and 932 have one pair of
servers — Server 0 and Server 1, one on each processor complex — both integrated in a
single Storage Facility Image (SFI). The DS8000 model 9B2 can have two pairs of servers
— two on each processor complex — each pair integrating one of the two possible SFIs.
You can issue the lsserver command to see the available servers.
Device Adapter (DA): A physical component of the DS8000 that provides communications
between the servers and the storage devices. The lsda command lists the available
device adapters.
Rank: An array site made into an array, which is then made into a rank. For the DS8000 a
rank is a collection of eight disk drive modules (DDMs). The lsrank command displays
detailed information about the ranks.
Additionally, it is always best to locate the FlashCopy target volume on the same DS8000
server as the FlashCopy source volume. It is also good practice to locate the FlashCopy
target volume on a different device adapter (DA) than the source volume, but there are some
cases where this is really an unimportant decision.
Tip: If the FlashCopy target volume is on the same rank as the FlashCopy source volume
the user runs the risk of a rank failure causing the loss of both the source and the target
volumes.
Tip: To find the relative location of your volumes, you can use the following procedure.
1. Use the lsfbvol command to learn which Extent Pool contains the relevant volumes.
2. Use the lsrank command to display both the device adapter and the rank for each
Extent Pool.
3. To determine which server contains your volumes, look at the Extent Pool name. Even
numbered Extent Pools are always from Server 0, while odd numbered Extent Pools
are always from Server1.
The LSSs/LCUs in a DS8000 disk subsystem are logical constructs that are no longer tied to
predetermined ranks. Within the DS8000 the LSS/LCU can be configured to span one or
more ranks but is not limited to specific ranks. There can be individual ranks that contain
volumes from more than a single LSS/LCU, which was not possible before the introduction of
the DS8000.
The FlashCopy establish phase, or logical FlashCopy, is the period of time when the
microcode is preparing things, such as the bitmaps, necessary to create the FlashCopy
relationship so the microcode can properly process subsequent reads and writes to the
related volumes. During this logical FlashCopy period, no writes are allowed to the volumes.
After the logical relationship has been established, normal I/O activity is allowed to both
source and target volumes according to the options selected.
When there are a large number of volumes, the method used to invoke the FlashCopy
commands can influence the time it takes to complete the logical FlashCopy for all FlashCopy
pairs. An in-band invocation will source the commands to the microcode faster than an
out-of-band method in most cases. An in-band establish is one that uses the CCW
commands directly, like in z/OS.
The fastest out-of-band method is the DS CLI. The method to control or invoke FlashCopy
should be selected based upon the total solution requirements and not strictly on logical
FlashCopy performance considerations unless this is a critical performance issue. For open
systems, the DS CLI is the fastest invocation method.
Finally, the placement of the FlashCopy source and target volumes has an effect on the
establish performance. You can refer to the previous section for a discussion on this topic as
well as to Table 9-1 on page 93 for a summary of the recommendations.
Note: The term collision describes a forced copy from the source to the target because a
write to the source has occurred. This occurs on the first write to a track. Note that since
the DS8000 writes to non-volatile cache, there is typically no direct response time delay on
host writes. The forced copy only occurs when the write is de-staged onto disk.
If a large number of volumes have been established, then do not expect to see all pairs
actively copying data as soon as their logical FlashCopy relationship is completed. The
DS8000 microcode has algorithms that will limit the number of active pairs copying data. This
algorithm will try to balance active copy pairs across the DS8000 device adapter resources.
Additionally, the algorithm will limit the number of active pairs such that there remains
bandwidth for host or server I/Os.
Tip: The DS8000 gives higher priority to application performance than background copy
performance. This means that the DS8000 will throttle the background copy if necessary,
so that applications are not unduly impacted.
The recommended placement of the FlashCopy source and target volumes, regarding the
physical FlashCopy phase, was discussed in the previous section. Refer to Table 9-1 on
page 93 for a summary of the conclusions. For the best background copy performance, the
implementation should always place the source and target volumes in different ranks. There
are additional criteria to consider if the FlashCopy is a full box copy that involves all ranks.
Note: The term full box copy has the implication that all rank resources are involved in the
copy process. Either all or nearly all ranks have both source and target volumes, or half the
ranks have source volumes and half the ranks have target volumes.
For full box copies, you should still place the source and target volumes in different ranks.
When all ranks are participating in the FlashCopy, it is still possible to accomplish this by
doing a FlashCopy of volumes on rank R0 onto rank R1, and volumes on rank R1 onto rank
R0 (for example). Additionally, if there is heavy application activity in the source rank,
performance would be less affected if the background copy target was in some other rank that
could be expected to have lighter application activity.
If background copy performance is of high importance in your environment, you should use
incremental FlashCopy as much as possible. Incremental FlashCopy will greatly reduce the
amount of data that needs to be copied, and therefore greatly reduces the background copy
time.
Note: As already mentioned, the recommendations discussed in this chapter only consider
the performance aspects of a FlashCopy implementation. But FlashCopy performance is
only one aspect of an intelligent system design. You must consider all business require-
ments when designing a total solution. These additional requirements, together with the
performance considerations, will guide you when choosing FlashCopy options like copy or
no copy and incremental, as well as when making choices on source and target volumes
location.
The relative placement of the source and target volumes has a significant impact on the
application’s performance, which has been discussed in 9.1.1, “Distribution of the workload -
source and target volumes location” on page 92.
Note: The incremental FlashCopy resyncflash command does not have a -nocp (no copy)
option. Using resyncflash will automatically use the copy option, regardless of whether the
original FlashCopy was copy or no copy.
The copy approach causes an initial copy of all source volumes, but then will only copy
changed tracks when the Incremental Resync command is invoked. The no copy approach
does not perform an initial copy of the source volumes when the Incremental Resync
command is invoked, but does copy on all first updates to source tracks since the preceding
FlashCopy. Again, this copy only occurs when that track is de-staged, so in many cases there
is no impact to application performance.
The designer needs to consider the consequences of bursts of background copy, initially and
following each resync (copy option), which might impact the application — versus continuous
impact, when there are collisions that require a source to target copy before the write update
can complete (no copy option).
If running incremental FlashCopy for an extended period of time, the copy option could very
well be the proper choice.
One method for protecting against failure is to use multiple FlashCopy targets. FlashCopy
supports up to 12 targets per source volume. With this feature, it is possible to keep up to 12
versions of your production data (for example, a FlashCopy backup every two hours for one
day). Another method is to use incremental FlashCopy. Incremental FlashCopy copies only
the data that has changed since the last FlashCopy, so the background copy completes much
faster.
The backup to tape is normally done shortly after the logical FlashCopy relationships have
been established for all of the volumes that are going to be backed up to tape.
If you choose the no copy option, that is probably because the data being backed up to tape
is mostly coming from the FlashCopy source volumes. If this is the case, then the location of
the target volumes is less critical and might be decided by considerations other than
performance.
If you choose the copy option, that is probably because the data being backed up is coming
from the target volumes — assuming that the backup to tape does not start until the
background copy completes. If the backup starts sooner, the data could be coming from a
mixture of source volumes and target volumes. As the backup continues, more and more of
the data will come from the target volumes as the background copy moves more and more of
the data to the target volumes.
To have the least impact on the application and to have a fast backup to tape, we recommend
that the source volumes be evenly spread across the available disk subsystems and the disk
subsystems resources. Once the backup to tape is complete, make sure to withdraw the
FlashCopy relationship.
Tip: Withdraw the pairs as soon as the backup to tape is finished. This eliminates any
additional copying from the source volume, either due to copy or collisions.
Tip: The recommended solution is to fully explore alternatives that would allow no
overlapping of FlashCopy activity with other peak application activity. If such alternatives
are not viable for whatever operative reasons, then consider the topics discussed in the
present section.
As discussed previously, the choice of whether to use copy or no copy depends mostly on
business requirements, but with regards to performance, it also depends a great deal on the
In general, no copy is the preferred method, but you should also think about the following
considerations when choosing either copy or no copy:
Using no copy. The argument here is that the impact caused by the I/O resulting from the
copy option is more significant than that of the no copy option, where less I/O activity
resulting from collisions occur. Still, since the background copy only occurs when the
writes are de-staged from non-volatile cache, there is typically negligible impact. If the
workload is cache friendly, then potentially all of the operations will be served from cache,
and there will be no impact from collisions at all.
Using copy. The goal of using copy is to quickly complete the background copy and hence
the overlapping situations between FlashCopy and application processing ends sooner. If
copy is used, then all I/Os experience some degradation as they compete for resources
with the background copy activity. However, this impact may be somewhat less than the
impact to the individual writes that a collision causes.
If FlashCopy no copy is active during a period of application high activity, there could be a
high rate of collisions (that is, de-stages being delayed so that the track image can be read
and then written to the FlashCopy target track to preserve the point-in-time copy). The
de-stage delay could cause degradation of the performance for all writes that occur during
the delay de-stage periods. Note that it is only the first write to a track that would cause a
collision, and only when that write gets de-staged. The reads do not suffer the collision
degradation.
If using the copy option, also consider these tips:
Examine the application environment for the highest activity volumes and the most
performance sensitive volumes.
Consider arranging the FlashCopy order such that the highest activity and most
performance sensitive volumes are copied early and the least active and least
performance sensitive volumes are copied last.
Tip: One approach to achieve a specified FlashCopy order would be to partition the
volmes into priority groups. Issue the appropriate FlashCopy commands for all volumes,
but use copy on only the highest priority group and no copy on all other groups. After a
specified period of time or after some observable event, issue FlashCopy commands to
the next highest priority group from no copy to copy. Continue in this manner until all vol-
umes are fully copied.
If a background copy is the desired end result and FlashCopy is to be started just before or
during a high activity period, consider the possibility of starting with no copy and converting to
copy after the high activity period has completed.
One might also want to examine the use of incremental FlashCopy in a high performance
sensitive activity period. Incremental FlashCopy automatically uses the copy option, so if the
no copy option was previously selected, using incremental FlashCopy may impact
performance by causing a full background copy.
If the incremental FlashCopy approach is chosen, it might be best to create a FlashCopy copy
relationship during a quiet time. To minimize the amount of data to be copied when taking the
desired point in time copy, schedule an incremental refresh sufficiently in advance of the
Such an approach has pros and cons. The disadvantage is the loss of 50% of the ranks for
normal application processing. The advantage is that FlashCopy writes to target volumes will
not compete against applications writing to the target volumes. This allows the background
copy to complete faster, and thus reduces the interference with application I/Os.
If planning a FlashCopy implementation at the disaster recovery (DR) site, two distinct
environments must be considered:
DR mirroring performance with and without FlashCopy active
Application performance if DR failover occurs
As it is assumed that with test systems or integration systems many write operations will
occur over the time period involved, we recommend doing a copy in the background
environment.
The application typically needs to be quiesced or briefly suspended before executing the
FlashCopy. Also, some applications cache their data, so this data may need to be flushed to
disk, using application methods, prior to running the FlashCopy (not covered in our example).
Running Part 1 and Part 2 (see Example 10-2) as one job would not work. The job would fail,
as volume 6101 could not be the target for one FlashCopy relationship and the source for
another FlashCopy relationship at the same time. You must wait until the background copy for
6100 and 6101 finishes successfully before starting Part 2.
Alternatively, you can also issue the rmflash command with the -cp and -wait parameters.
These parameters cause the command to wait until the background copy is complete before
continuing with the next step.
102 IBM System Storage DS8000 Series: Copy Services in Open Environments
#=== 03:00 pm 6100 → 6300
#--- after the full volume copy of 6100 → 6101 finished
#--- establish relationship from 6101:6300
mkflash -dev IBM.2107-7506551 -seqnum 02 6101:6300
lsflash -dev IBM.2107-7506551 -l 6100-6300
Whenever the test environment needs to be reset to the original data, just run Part 2 of the
scripts or use the DS SM front-end to perform a FlashCopy.
The following are the steps involved in this procedure (refer Example 10-3 and
Example 10-4):
1. Part 1: Establish FlashCopy volume A → volume B with the no copy option.
2. Run backup.
3. Part 2: Remove the FlashCopy relationship once volume backup completes.
After the backup has been taken, remove the FlashCopy relationship if you don’t intend to use
it for other purposes. Thus, unnecessary writes can be avoided. See Example 10-4.
Doing backup based on the target volume allows you to use a target multiple times.
In complex application environments (for example, SAP), FlashCopy is often used as part of
the backup solutions. Good examples for such solutions are the IBM Tivoli® Storage Manager
for Hardware modules, which integrate into the IBM Tivoli Storage Manager backup
infrastructure.
Example 10-5 Create an initial FlashCopy ready for subsequent incremental FlashCopies
#=== Part 1: establish FlashCopy relationship
#--- remove existing FlashCopy relationships for volume 6100
rmflash -dev IBM.2107-7506551 -quiet 6100:6300
After the initial full volume copy, the following script supports the incremental copy of the
FlashCopy relationship, see Example 10-6.
10.2.3 Using a target volume to restore its contents back to the source
Logs may need to be applied to the target, and then the target volume is reversed to the
source volume.
For each source volume, one FlashCopy can exist with the -record and -persist attributes
set. Use this volume to refresh the source volume. To reverse the relationship, the data must
have been copied completely to the target before reversing it back to the source. To avoid a
situation where the full volume needs to be copied with each FlashCopy, the Incremental
FlashCopy should be used. Since logs may need to be applied to the target volume prior to
reversing it, the target volume should be write-enabled.
104 IBM System Storage DS8000 Series: Copy Services in Open Environments
After the initial FlashCopy, the incremental copies can be done; see Example 10-8.
The reverse of the FlashCopy is done using the reverseflash command; see Example 10-9.
This part ends with examples of Metro Mirror management and setup.
Note: Throughout this part of the book, in our discussions of Metro Mirror in open systems
environments, you will find that the term volume is used indistinctly from the term LUN. In
fact, you will see that the term volume is almost always used.
As data is synchronously transferred, the distance between the local and the remote disk
subsystems will determine the effect on application response time. Figure 11-1 illustrates the
sequence of a write update with Metro Mirror.
4
Server
Write acknowledge
write
1 Write to secondary
2
LUN or 3 LUN or
volume volume
Write complete
Primary Secondary
(source)
acknowledgment (target)
When the application performs a write update operation to a source volume, this is what
happens:
1. Write to source volume (DS8000 cache and NVS).
2. Write to target volume (DS8000 cache and NVS).
3. Signal write complete from the remote target DS8000.
4. Post I/O complete to host server.
The Fibre Channel connection between the local and the remote disk subsystems can be
direct, through a switch, or through other supported distance solutions (for example, Dense
Wave Division Multiplexor or DWDM).
110 IBM System Storage DS8000 Series: Copy Services in Open Environments
11.2 Metro Mirror volume state
Volumes participating in a Metro Mirror session can be found in either of the following states:
Copy pending: Volumes are in copy pending state after the Metro Mirror relationship is
established, but the source and target volumes are still out-of-sync. In that case, data still
needs to be copied from the source to the target volume of the Metro Mirror pair. This may
be the case immediately after a relationship is initially established, or re-established after
being suspended. The Metro Mirror target volume is not accessible when the pair is in
copy pending state.
Full duplex: The full duplex state is the state of a volume copy pair whose members are in
sync, that is, both source and target volumes contain exactly the same data. The target
volume is not accessible when the pair is in full duplex.
Suspended: Volumes are in the suspended state when the source and target storage
subsystems cannot communicate anymore, or when the Metro Mirror pair is suspended
manually. In this state, writes to the source volume are not mirrored onto the target
volume. The target volume becomes out-of-sync. During this time, Metro Mirror keeps a
bitmap record of the changed tracks in the source volume. Later, when the volumes are
re-synchronized, only the tracks that were updated will be copied.
Target copy pending: This indicates that the source volume is unknown or cannot be
queried and the target state is copy pending.
Target full-duplex: This indicates that the source volume is unknown or cannot be queried
and the target state is full duplex.
Target suspended: This indicates that the source volume is unknown or cannot be queried
and the target state is suspended.
Not remote copy pair: This indicates that the relationship is not a Metro Mirror pair.
Invalid-state: This indicates that the relationship state is invalid.
For Metro Mirror, consistency requirements are managed through use of the Consistency
Group option. You can specify this option when you are defining Metro Mirror paths between
pairs of LSSs or when you change the default LSS settings. Volumes or LUNs that are paired
between two LSSs whose paths are defined with the Consistency Group option can be
considered part of a Consistency Group.
Consistency is provided by means of the extended long busy (for z/OS) or queue full (for open
systems) conditions. These are triggered when the DS8000 detects a condition where it
cannot update the Metro Mirror target volume. The volume pair that first detects the error will
go into the queue full condition, such that it will not do any writes, and an SNMP trap message
will be issued. These messages can be used as a trigger for automation purposes that will
provide data consistency.
Data consistency and dependent writes are discussed in detail in 12.4, “Consistency Group
function” on page 116.
Metro Mirror itself does not offer a means of controlling this scenario — it offers the
Consistency Group and Critical attributes, which along with appropriate automation solutions
can manage data consistency and integrity at the remote site. The Metro Mirror volume pairs
are always consistent, due to the synchronous nature of Metro Mirror. However,
cross-volume or LSS data consistency must have an external management method. IBM
offers the GDPS® and eRCMF service offerings to deliver solutions in this area. Visit the IBM
Web site and see the Services and Industry Solutions page for more information.
When planning a Metro Mirror environment, the following topics should be considered:
Design of the automation solution
Maintenance
Testing
Support
All these must be considered. You do not want to be in a situation where you are relying on
your mirroring implementation for data to be consistent in a disaster situation, only to find that
it has not worked, or perhaps worse, you not being aware that your data is not consistent.
IBM offers services and solutions for the automation and management of the Metro Mirror
environment. These include GDPS and eRCMF. Visit the IBM Web site and see the Services
and Industry Solutions page for more information.
There are also solutions available for integrating Metro Mirror into a cluster, like HACMP/XD
for AIX, or IBM Total Storage Continuous Availability for Windows (GDS). See Part 9,
“Solutions” on page 537.
112 IBM System Storage DS8000 Series: Copy Services in Open Environments
12
When establishing a Metro Mirror pair, the following additional options can be selected:
No copy
This option does not copy data from the source to the target. This option presumes that
the volumes are already synchronized. The data synchronization is your responsibility and
the DS8000 does not check its synchronization.
Target read
This option allows host servers to read from the target. For a host server to read from the
target, the pair (source-target) must be in a full duplex state. This parameter applies to
open systems volumes (it does not apply to System z volumes.
In the open system’s file system environment, even if an application reads data from a file
system, a SCSI write command may be issued to the LUN where the file system resides.
This is because some information, such as the last access time stamp, may be updated by
the file system. In this case, even if you specify this option, the operation may fail.
Suspend after data synchronization
This option suspends the volume pairs after the data has been synchronized.
Wait option
This option delays the command response until the volume pairs are in one of the final
states: simplex, full duplex, suspended, target full duplex, target suspended (until the pair
is not in the copy pending state). This parameter cannot be used with -type gcp or -mode
nocp.
Reset reserve on target
This option allows the establishment of a Metro Mirror relationship when the target volume
is reserved by another host. This parameter can only be used with open system volumes.
If this option is not specified and the target volume is reserved, the command fails.
114 IBM System Storage DS8000 Series: Copy Services in Open Environments
Terminate Metro Mirror pair
This operation ends the Metro MIrror relationship between the source and target volumes.
12.2 Clustering
Since the disk attached to the server is being mirrored using Metro Mirror, this offers
improved opportunities for high availability solutions.
There are also solutions available for integrating Metro Mirror into a cluster, like HACMP/XD
for AIX, and IBM Total Storage Continuous Availability for Windows (GDS). For more
information refer to Part 9, “Solutions” on page 537.
In a typical Metro Mirror environment, processing will temporarily switch over to the Metro
Mirror remote site upon an outage at the local site. When the local site is capable of resuming
production, processing will switch back from the remote site to the local site.
At the recovery site, the Metro Mirror Failover function combines into a single task the three
steps involved in the switchover (planned or unplanned) to the remote site: Terminate the
original Metro Mirror relationship, then establish and suspend a new relationship at the
remote site. The state of the original source volume at the normal production site is
preserved. The state of the original target volume at the recovery site becomes a source
suspended. This design takes into account that the original source LSS may no longer be
reachable.
To initiate the switchback to the production site, the Metro Mirror Failback function, at the
recovery site, checks the preserved state of the original source volume at the production site
to determine how much data to copy back. Then either all tracks or only out-of-sync tracks
are copied, with the original source volume becoming a target full-duplex. In more detail, this
is how Metro Mirror Failback operates:
If a volume at the production site is in the simplex state, all of the data for that volume is
copied back from the recovery site to the production site.
If a volume at the production site is in the full-duplex or suspended state and without
changed tracks, only the modified data on the volume at the recovery site is copied back to
the volume at the production site.
If a volume at the production site is in a suspended state and has tracks that have been
updated, then both the tracks changed on the production site and the tracks marked at the
recovery site will be copied back.
Finally, the volume at the production site becomes a write-inhibited target volume. This
action is performed on an individual volume basis.
In this manner, Metro Mirror Failover and Failback are dual operations. It is possible to
implement all site switch operations with two pairs of failover and failback tasks (one pair for
each direction).
Note: For a planned switchover from site (A) to site (B), and in order to keep data consis-
tency at (B), the application at site (A) has to be quiesced before the Metro Mirror Failover
operation at (B). Alternatively, you can use the freezepprc and unfreezepprc commands.
The same consideration applies when switching back from site (B) to (A).
The Metro Mirror Failover and Failback modes can be invoked from the DS Storage Manager
GUI or the DS Command-Line Interface (CLI).
We give an example of a Metro Mirror Failover and Failback sequence and how to control it in
14.8, “Failover and Failback functions for sites switching” on page 147.
116 IBM System Storage DS8000 Series: Copy Services in Open Environments
a certain procedure is necessary to keep data consistency even in a synchronous remote
copy environment.
In this section we explain data consistency and explain how the Metro Mirror Consistency
Group function keeps data consistency at the remote site in case of a rolling disaster.
If the copy of the data contains any of these combinations then the data is consistent:
Operation 1, 2, and 3
Operation 1 and 2
Operation 1
If the copy of data contains any of those combinations then the data is inconsistent (the order
of dependent writes was not preserved):
Operation 2 and 3
Operation 1 and 3
Operation 2
Operation 3
When discussing the Consistency Group function, data consistency means this sequence is
always kept in the copied data. And the order of non-dependent writes does not necessarily
need to be preserved. For example, consider the following two sequences:
1. Deposit paycheck in checking account A
2. Withdraw cash from checking account A
3. Deposit paycheck in checking account B
4. Withdraw cash from checking account B
In order for the data to be consistent, the deposit of the paycheck must be applied before the
withdraw of cash for each of the checking accounts. However, it does not matter whether the
deposit to checking account A or checking account B occurred first, as long as the associated
withdrawals are in the correct order. So for example, the data copy would be consistent if the
following sequence occurred at the copy.
1. Deposit paycheck in checking account B
2. Deposit paycheck in checking account A
3. Withdraw cash from checking account B
4. WIthdraw cash from checking account A
In other words, the order of updates is not the same as it was for the source data, but the
order of dependent writes is still preserved.
In the disk subsystem itself, each command is managed with each logical subsystem (LSS).
This means that there are slight time lags until each volume in the different LSS is subject to
the queue full condition. Some people are concerned that the time lag causes you to lose
data consistency, but that is not true. We explain how to keep data consistency in the
Consistency Group environments in the following section.
In the example in Figure 12-2, three write operations (first, second, and third) are dependent
writes. This means that these operations must be completed sequentially.
LSS11 LSS21
1st
Wait
Application
on
LSS12 LSS22
Servers
2nd
Wait
LSS13 LSS23
3rd
Wait
dependent write
operation
Source DS8000 Target DS8000
Figure 12-2 Consistency Group: Example1
In Figure 12-2, there are two Metro Mirror paths between LSS11 and LSS21. There are
another two Metro Mirror paths for each of the other LSS pairs (LSS12;LSS22 and
LSS13:LSS23). In a disaster, the paths may fail at different times. At the beginning of the
disaster, such as a rolling disaster, one set of paths (such as the paths between LSS11 and
118 IBM System Storage DS8000 Series: Copy Services in Open Environments
LSS21) may be inoperable while other paths are working. At this time, the volumes in LSS11
are in a queue full condition, and the volumes in LSS12 and 13 are not. The first operation is
not completed because of the queue full condition, and the second and third operations are
not completed because the first operation has not been completed. In this case, the first,
second, and third updates are not included in the Metro Mirror target volumes in LSS21,
LSS22, and LSS23. Therefore, the Metro Mirror target volumes at the remote site keep
consistent data.
In the example illustrated in Figure 12-3, the volumes in LSS12 are in a queue full condition
and the other volumes in LSS11 and 13 are not. The first write operation is completed
because the volumes in LSS11 are not in an queue full condition. The second write operation
is not completed because of the queue full condition. The third write operation is also not
completed because the second operation is not completed. In this case, the first update is
included in the Metro Mirror target volumes, and the second and third updates are not
included. Therefore, this case is also consistent.
LSS11 LSS21
1st
Completed
Application
on
LSS12 LSS22
Servers
2nd
Wait
LSS13 LSS23
3rd
Wait
dependent write
operation
Source DS8000 Target DS8000
Figure 12-3 Consistency Group: Example 2
In all cases, if each write operation is dependent, the Consistency Group option can keep the
data consistent in the Metro Mirror target volumes until the Consistency Group time-out
occurs. After the time-out value has been exceeded, all held I/O will be released. The
Consistency Group time-out value can be specified at the LSS level.
LSS11 LSS21
1st
Completed
Application
on
LSS12 LSS22
Servers
2nd
Wait
LSS13 LSS23
3rd
Wait Completed
independent
write Source DS8000 Target DS8000
operations
In this case, the Metro Mirror target volumes reflect only the first and third write operations,
not the second operation.
Typical database management software can recover its databases by using its log files if
dependent write operations are kept. At the same time you should not need to care whether
independent write operations are kept in sequence.
When the DS8000 detects a condition where it cannot update the Metro Mirror target volume,
the Metro Mirror source volume with the Consistency Group option becomes suspended and
enters the queue full condition. At the same time the DS8000 issues an SNMP notification
(Trap 202: source PPRC Devices on LSS Suspended Due to Error).
120 IBM System Storage DS8000 Series: Copy Services in Open Environments
An automation program, triggered by the SNMP notification, can issue the freezepprc
command to all LSS pairs that have volumes related to the application. This command
causes all Metro Mirror source volumes in the LSSs with the Consistency Group option to
become suspended and enter the queue full condition. In addition, this operation removes the
Metro Mirror paths. Because all Metro Mirror source volumes become suspended and all
related paths are removed, further updates to the source volumes will not be sent to their
targets. Now you have consistent data at the remote site. It is necessary to have an
automation procedure like the one just described in order to have consistent data at the
remote site in case of a rolling disaster.
In case of partial link failures (such as in the case of Figure 12-2 on page 118), and in order to
rapidly resume application I/O processing at the local site, you can use the unfreezepprc
command to resume held I/Os to the affected LSSs. In general, you will issue the
unfreezepprc command after successful completion of the freezepprc command to all
related LSSs, which means that you have consistent data at the remote site.
Note: By partial links failure we mean (in this section) the situation where for a particular
LSS all of its links fail, while at the same time there are other LSSs with their links working.
When any particular LSS has part of its links not working, Metro Mirror can keep sending
data to the target volumes using the other available links in the LSS.
The default two-minute timer of the queue full condition gives the automation enough time to
issue a freezepprc command to the necessary LSSs. I/O resumes after the default
two-minute interval if a unfreezepprc command is not received first.
It is not possible to issue Consistency Group (freeze/unfreeze) type commands from the DS
Storage Manager GUI, but you can change the time-out values by using the DS GUI.
Important: The queue full condition is presented only for the source volume affected by
the error (in the case of path failures, multiple volumes will often be affected). Still, the
freeze operation is performed at the LSS level, causing all Metro Mirror volumes in that
LSS to go into suspended state with queue full condition and terminating all associated
paths. Therefore, when planning your implementation, you should consider not intermixing
volumes from different applications in an LSS pair that is part of a Consistency Group.
Otherwise the not-in-error volumes belonging to other applications will be frozen, too.
Note: For Metro Mirror, the DS8000 supports Fibre Channel links only. To facilitate ease of
testing, the DS8000 does support Metro Mirror source and target on the same DS8000.
LSS 0 LSS 0
LSS 1 Physical LSS 1
LSS 2 Fibre Channel link LSS 2
LSS 3 LSS 3
: :
LSS 08 256 logical paths LSS 08
: per FCP link :
LSS nn LSS nn
Figure 12-5 Logical paths
Paths are unidirectional, that is, they are defined to operate in either one direction or the
other. Still, Metro Mirror is bidirectional. That is, any particular pair of LSSs can have paths
defined among them that have opposite directions. Each LSS holds both source and target
volumes from the other particular LSS. Also, opposite direction paths are allowed to be
defined on the same Fibre Channel physical link.
For bandwidth and redundancy, more than one path can be created between the same LSSs.
Metro Mirror will balance the workload across the available paths between the source and
target LSSs.
Note: Remember that the LSS is not a physical construct in the DS8000 — it is a logical
construct. Volumes in an LSS can come from multiple disk arrays.
Physical links are bidirectional and can be shared by other Metro Mirror pairs as well as other
remote mirror and copy functions, such as Global Copy and Global Mirror.
122 IBM System Storage DS8000 Series: Copy Services in Open Environments
Although one FCP link would have sufficient bandwidth for most Metro Mirror environments,
for redundancy reasons we recommend configuring two Fibre Channel links between each
source and remote disk subsystem.
Dedicating Fibre Channel ports for Metro Mirror use guarantees no interference from host I/O
activity. This is recommended with Metro Mirror, which is time critical and should not be
impacted by host I/O activity. Each Metro Mirror port provides connectivity for all LSSs within
the DS8000 and can carry multiple logical Metro Mirror paths.
Note: If you want the same set of Fibre Channel links to be shared between Metro Mirror
and Global Copy or Global Mirror, you should consider the impact of the aggregate data
transfer. In general, we do not recommend sharing the FCP links used for Metro Mirror,
including the physical network links, with other asynchronous remote copy functions.
Metro Mirror FCP links can be directly connected, or connected by up to two switches.
Note: If you use channel extension technology devices for Metro Mirror links, you should
verify with the product’s vendor what environment (directly connected or connected with a
SAN switch) is supported by the vendor and what SAN switch is supported.
Figure 12-6 shows an example where we have a 1:1 mapping of source to target LSSs, and
where the three logical paths are accommodated in over one physical link:
LSS1 in DS8000-1 to LSS1 in DS8000-2
LSS2 in DS8000-1 to LSS2 in DS8000-2
LSS3 in DS8000-1 to LSS3 in DS8000-2
Alternatively, if the volumes in each of the LSSs of DS8000-1 map to volumes in all three
target LSSs in DS8000-2, there will be nine logical paths over the physical link (not fully
illustrated in Figure 12-6). Note that we recommend a 1:1 LSS mapping.
DS8000 1 DS8000 2
3-9 logical paths
LSS 1 LSS 1
LSS 3 LSS 3
Metro Mirror
paths
1 logical path 1 logical path
Figure 12-6 Logical paths over a physical link for Metro Mirror
12.6 Bandwidth
Prior to establishing your Metro Mirror solution, you should establish what your peak
bandwidth requirement will be. This will help to ensure that you have enough Metro Mirror
links in place to support that requirement.
To avoid any response time issues you should establish the peak write rate for your systems
and ensure that you have adequate bandwidth to cope with this load, and allow for growth.
Remember that only writes are mirrored across to the target volumes.
Tools to assist with this are operating system dependent, like iostat. Refer to the IBM
Redbook The IBM TotalStorage DS8000 Series: Implementation, SG24-6786. Another
method, but not so exact, is to monitor the traffic over the FC switches using FC switch tools
and other management tools, and remember that only writes will be mirrored by Metro Mirror.
You can also get some feeling about the proportion of read/writes by issuing datapath query
devstats on SDD-attached servers.
This can also simplify the Metro Mirror environment, as it is possible to reduce the number of
commands that are required for data consistency as well as make its effects more granular.
For example, a freeze operation is performed at the LSS level, causing all Metro Mirror
volumes in that LSS to go into suspended state with a queue full condition and terminate all
associated paths. If you assign LSSs to each of your applications, you can control the impact
of the queue full condition caused by the freeze operation at an application level. On the
contrary, if you put volumes used by several different applications into the same LSS, all those
applications sharing the LSS and their Metro Mirror volumes will be affected by the queue full
condition. You can issue one freezepprc command to multiple LSSs. Therefore, you do not
need to consolidate the number of LSSs that your applications use. See 12.4.2, “Consistency
Group function - how it works” on page 118 for more information.
124 IBM System Storage DS8000 Series: Copy Services in Open Environments
12.8 Distance
The distance between your local and remote DS8000 subsystems has an effect on the
response time overhead of the Metro Mirror implementation. The maximum supported
distance for Metro Mirror is 300 km.
Figure 12-7 on page 126 shows this idea in a graphical form. DS8000 #1 has Metro Mirror
paths defined to DS8000 #2, which is in a remote location. On DS8000 #1, volumes defined
in LSS 00 are mirrored to volumes in LSS 00 on DS8000 #2 (volume P1 is paired with volume
S1, P2 with S2, P3 with S3, and so on). Volumes in LSS 01 on DS8000 #1 are mirrored to
volumes in LSS 01 on DS8000 #2, and so on. Requirements for additional capacity can be
added in a symmetrical way also by addition of volumes into existing LSSs, and by the
addition of new LSSs when needed. (For example, addition of two volumes in LSS 03 and 05,
and one volume to LSS 04 will bring them to the same number of volumes as the other LSSs.
Additional volumes could then be distributed evenly across all LSSs, or additional LSSs
added.)
12.10 Volumes
You need to consider which volumes should be mirrored to the target site. One option is to
mirror all volumes. This is advantageous for the following reasons:
You will not need to consider whether any required data has been missed or not.
Users will not need to remember which logical pool of volumes is mirrored and which is
not.
Adding volumes to the environment is simplified. You do not have to have two processes
for addition of disk (one for mirrored volumes, and another for non-mirrored volumes).
You will be able to move data around your disk environment easily without worrying about
whether the target volume is a mirrored volume.
You may choose not to mirror all volumes. In this case, you will need careful control over what
data is placed on the mirrored volumes (to avoid any capacity issues) and what data you
place on the non-mirrored volumes (to avoid missing any required data). One method of
doing this is to place all mirrored volumes in a particular set of LSSs, in which all volumes are
Metro Mirror-enabled, and direct all data requiring mirroring to these volumes.
Though mirroring all volumes might be the simpler solution to manage, it could also require
significantly more network bandwidth. Since network bandwidth is a cost to the solution,
minimizing the bandwidth may well be worth the added management complexity.
126 IBM System Storage DS8000 Series: Copy Services in Open Environments
12.11 Hardware requirements
Metro Mirror is an optional licensed function available on the DS8000. Licensed functions
require the selection of a DS8000 series feature number (IBM 2107) and the acquisition of
DS8000 series Function Authorization (IBM 2244) feature numbers:
The 2107 licensed function indicator feature number enables the technical activation of the
function subject to the client applying a feature activation code made available by IBM.
The 2244 function authorization feature numbers establish the extent of IBM authorization
for that function on the 2107 machine for which it was acquired.
To use Metro Mirror you must have the 2107 function indicator feature (#0744) and the
corresponding DS8000 Series Function Authorization (IBM 2244-RMC) with the adequate
feature number (#7440-7450) to establish the extent of IBM authorization: license size in
terms of physical capacity. The DS8000 Series Function Authorizations are for billing
purposes only and establish the extent of IBM authorization for use of a particular licensed
function on the IBM System Storage DS8000 series (IBM 2107). Authorization must be
acquired for both the source and target 2107 machine.
Note: For a detailed explanation of the features involved and the considerations you must
have when ordering Metro Mirror we recommend that you refer to the announcement
letters:
IBM System Storage DS8000 — Functions Authorization (IBM 2244).
IBM System Storage DS8000 Series (IBM 2107)
Interoperability
Metro Mirror pairs can only be established between disk subsystems of the same (or similar)
type and features. For example, a DS8000 can have a Metro Mirror pair with another DS8000,
a DS6000, an ESS 800, or an ESS 750. It cannot have a Metro Mirror pair with an RVA or an
ESS F20. Note that all disk subsystems must have the appropriate Metro Mirror feature
installed. If your DS8000 is being mirrored to an ESS disk subsystem, the ESS must have
PPRC Version 2 (which supports Fibre Channel links) with the appropriate licensed internal
code level (LIC).
Bandwidth analysis and capacity planning for your Metro Mirror links should help to define
how many links you need, and when you need to add more, to ensure best possible
performance.
As part of your implementation project, you may be able to identify and then distribute hot
spots across your configuration, or take other actions to manage and balance the load.
As an alternative, you may also choose to do the initial synchronization in the Global Copy
mode, and then switch to the Metro Mirror mode. This will allow you to bring all volumes to
duplex-pending XD state, which has little or no application impact, and then switch to full
duplex state. You can do so using the DS CLI.
13.2 Scalability
The DS8000 Metro Mirror environment can be scaled up or down as required. If new volumes
are added to the DS8000 that require mirroring, they can be dynamically added. If additional
Metro Mirror paths are required, they also can be dynamically added.
As we have previously mentioned, the logical nature of the LSS has made a Metro Mirror
implementation on the DS8000 easier to plan, implement, and manage. However, if you need
to add more LSSs to your Metro Mirror environment, your management and automation
solutions should be set up to handle this.
130 IBM System Storage DS8000 Series: Copy Services in Open Environments
The eRCMF and GDPS service offerings are designed to provide this functionality. Visit the
IBM Web site and see the Services and Industry Solutions page for more information. Also
see Part 9, “Solutions” on page 537.
The present chapter gives an overview of the DS CLI and the DS SM for Metro Mirror
management. DS CLI and DS SM are also covered in Part 2, “Interfaces” on page 15.
eRCMF is a scalable and flexible solution to protect business data and maintain consistent
data, providing an automation tool that simplifies the disaster recovery operations. For more
information about this solution see Part 9, “Solutions” on page 537.
Also the following interfaces, not covered in this book, can be used:
DS Open Application Programming Interface (DS Open API)
z/OS interfaces: TSO, ICKDSF, and ANTRQST API
In order to use a z/OS interface to manage open systems LUNs, the DS8000 must have at
least one CKD volume. If you are interested in this possibility, refer to the IBM Redbook The
IBM TotalStorage DS8000 Series: Copy Services with IBM Eserver zSeries, SG24-6787.
For information about the DS Open API, refer to the publication IBM System Storage DS
Open Application Programming Interface Reference, GC35-0516.
List available I/O ports that can be lsavailpprcport This information is shown
used to establish Metro Mirror during the process when a
paths. path is established.
134 IBM System Storage DS8000 Series: Copy Services in Open Environments
Task Command with DS CLI Select option with DS SM
The DS CLI has the advantage that you can make scripts and use them for automation. The
DS Storage Manager GUI is a web-based graphical user interface and is more intuitive than
the DS CLI.
Customer
Network DS HMC 1
Ethernet Processor
(internal) Complex 0
DS Storage
Manager
Switch 1
DS Storage
Manager realtime
DS CLI DS8000
DS API
(DS HMC 2) Ethernet Processor
(external) Switch 2 Complex 1
DS Storage
Manager realtime
mark
mark
mark
Figure 14-1 DS8000 Copy Services network components
Each DS8000 will have an internal DS HMC in the base frame, and you can have an external
DS HMC for redundancy.
You need at least one available DS HMC to issue Copy Services commands. If you have only
one DS HMC and if it has a failure, you will not be able to issue Copy Services commands.
Therefore, we recommend having a dual DS HMC configuration, as in Figure 14-1 on
page 135, especially when using automation scripts to run Copy Services functions so that
the script keeps working in case of one DS HMC failure.
To establish a Metro Mirror pair, the LAN connection between the source and target DS8000s
is not mandatory, unlike the ESS. A similar situation happens with Global Copy. However,
copy services DS CLI commands have to be issued to the DS HMC connected to the
DS8000, which has the source volume. When you need to establish a Metro Mirror pair from
a volume at the remote site to a volume at the local site, the DS CLI command has to be
issued to the DS8000 HMC at the remote site. In this case, you need a LAN connection from
the local DS CLI client machine to the DS8000 at the remote site.
When you establish or remove Metro Mirror paths and volume pairs, you must give the DS
CLI commands to the DS HMC that is connected to the source DS8000. Also, when checking
status information at the local and remote site, you must issue DS CLI list type commands,
such as the lspprc, to each DS8000, source, and target.
The DS CLI commands are documented in the publication IBM System Storage DS8000
Command-Line Interface User´s Guide, SC26-7916.
For a detailed explanation of the DS SM Graphical User Interface, refer to Chapter 3, “DS
Storage Manager (GUI)” on page 17.
Note: The DS SM GUI supports (differently from the DS CLI) a multi-session environment,
but not all functions and options are supported by the GUI.
136 IBM System Storage DS8000 Series: Copy Services in Open Environments
14.5 Set up Metro Mirror environment using DS CLI
In the following sections we present an example of how to set up a Metro Mirror environment
using the DS CLI. Figure 14-2 shows the configuration that we implement.
LSS10 LSS20
1000 2000
1001 2001
1100 2100
1101 2101
DS8000#1 DS8000#2
-dev IBM.2107-7520781 -dev IBM.2107-75ABTV1
Figure 14-2 DS8000 configuration in the Metro Mirror setup example
In our example we use different LSS and LUN numbers for the Metro Mirror source and target
elements, so that you can more clearly understand which one is being specified when going
through the reading of the example.
Note: In a real environment (and different from our example), to simplify the management
of your Metro Mirror environment, we recommend that you maintain a symmetrical configu-
ration in terms of both physical and logical elements.
To save you from typing -dev storage_image_ID and -remotedev storage_image_ID in each
command, you can put these values in your CLI profile. The default and target Storage Image
ID devid and remotedevid are equivalent to the -dev storage_image_ID and -remotedev
storage_image_ID command options. See Example 14-1.
Example 14-1 Part of CLI profile
# Internal SHMC ipaddress for DS8000#1
hmc1: 10.10.10.1
# External SHMC ipaddress for DS8000#1
hmc2: 10.10.10.2
# Default and target Storage Image ID
devid: IBM.2107-7520781
remotedevid:IBM.2107-75ABTV1
For further information see Chapter 4, “DS Command-Line Interface” on page 23.
dscli>
1000 2000
Metro Mirror Pairs
1001 2001
1100 2100
Metro Mirror Pairs
1101 2101
DS8000#1 DS8000#2
-dev IBM.2107-7520781 -dev IBM.2107-75ABTV1
138 IBM System Storage DS8000 Series: Copy Services in Open Environments
14.5.3 Determine the available Fibre Channel links
First you must look at the available Fibre Channel links. With the lsavailpprcport command,
you can do this (see Example 14-3). You see all available port combinations between the
source and the target LSSs. You have to issue this command to the DS HMC connected to
DS8000#1, which is the Metro Mirror source.
FCP port ID with the lsavailpprcport command has four hexadecimal characters in the
format 0xEEAP, where EE is a port enclosure number (00–3F), A is the adapter number
(0–F), and P is the port number (0–F). The FCP port ID number is prefixed with the letter I.
You can use the -fullid parameter to display the DS8000 Storage Image ID in the command
output (see Example 14-4).
Example 14-4 List available fibre links with DS8000 Storage Image ID
dscli> lsavailpprcport -l -fullid -remotedev IBM.2107-75ABTV1 -remotewwnn 5005076303FFC663 10:20
Date/Time: October 25, 2005 10:00:26 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
Local Port Attached Port Type Switch ID Switch Port
========================================================================
IBM.2107-7520781/I0143 IBM.2107-75ABTV1/I0010 FCP NA NA
IBM.2107-7520781/I0213 IBM.2107-75ABTV1/I0140 FCP NA NA
You need the worldwide node name (WWNN) of your target DS8000 to issue the
lsavailpprcport command. You can get this by using the lssi command; see Example 14-5.
You have to issue this command to the DS HMC connected to DS8000#2, which is the Metro
Mirror target.
With the lspprcpath command you can use the -fullid command flag to display the fully
qualified DS8000 Storage Image ID in the command output; see Example 14-7.
Example 14-8 Create Metro Mirror pairs and verify the result
dscli> mkpprc -remotedev IBM.2107-75ABTV1 -type mmir 1000-1001:2000-2001 1100-1101:2100-2101
Date/Time: October 25, 2005 11:19:06 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 1000:2000 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 1001:2001 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 1100:2100 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 1101:2101 successfully created.
dscli> lspprc 1000-1001 1100-1101
Date/Time: October 25, 2005 11:26:59 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
===================================================================================================
1000:2000 Copy Pending - Metro Mirror 10 unknown Disabled Invalid
1001:2001 Copy Pending - Metro Mirror 10 unknown Disabled Invalid
1100:2100 Copy Pending - Metro Mirror 11 unknown Disabled Invalid
1101:2101 Copy Pending - Metro Mirror 11 unknown Disabled Invalid
Once the Metro Mirror source and target volumes have been synchronized, the volume state
changes to Full Duplex from Copy Pending; see Example 14-9.
Example 14-9 List Metro Mirror status after Metro Mirror initial copy completes
dscli> lspprc 1000-1001 1100-1101
Date/Time: October 25, 2005 11:28:55 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
==================================================================================================
1000:2000 Full Duplex - Metro Mirror 10 unknown Disabled Invalid
140 IBM System Storage DS8000 Series: Copy Services in Open Environments
1001:2001 Full Duplex - Metro Mirror 10 unknown Disabled Invalid
1100:2100 Full Duplex - Metro Mirror 11 unknown Disabled Invalid
1101:2101 Full Duplex - Metro Mirror 11 unknown Disabled Invalid
The states of full duplex and copy pending indicate the Metro Mirror source state. In the case
of the target state, the states are Target Full Duplex and Target Copy Pending; see
Example 14-10. You have to give this command to the DS HMC connected to DS8000#2,
which is the Metro Mirror target.
In the copy pending state, you can check the data transfer status of the Metro Mirror initial
copy by using the lspprc -l command. Out Of Sync Tracks shows the remaining tracks to be
sent to the target volume. The size of the logical track for FB volume for the DS8000 is 64 KB.
And you also can use the lspprc -fullid command to show the DS8000 Storage Image ID
in the command output; see Example 14-11.
Example 14-11 lspprc -l and lspprc -fullid for Metro Mirror pairs
dscli> lspprc -l 1000-1001 1100-1101
Date/Time: October 25, 2005 11:26:22 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID State Reason Type Out Of Sync Tracks Tgt Read Src Cascade Tgt Cascade Date Suspended SourceLSS Timeout (secs)
Critical Mode First Pass Status
==========================================================================================================================================
1000:2000 Copy Pending - Metro Mirror 46725 Disabled Disabled invalid - 10 unknown
Disabled Invalid
1001:2001 Copy Pending - Metro Mirror 46579 Disabled Disabled invalid - 10 unknown
Disabled Invalid
1100:2100 Copy Pending - Metro Mirror 44080 Disabled Disabled invalid - 11 unknown
Disabled Invalid
1101:2101 Copy Pending - Metro Mirror 44040 Disabled Disabled invalid - 11 unknown
Disabled Invalid
dscli> lspprc -fullid 1000-1001 1100-1101
Date/Time: October 25, 2005 11:28:18 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass
Status
==========================================================================================================================================
IBM.2107-7520781/1000:IBM.2107-75ABTV1/2000 Full Duplex - Metro Mirror IBM.2107-7520781/10 unknown Disabled Invalid
IBM.2107-7520781/1001:IBM.2107-75ABTV1/2001 Full Duplex - Metro Mirror IBM.2107-7520781/10 unknown Disabled Invalid
IBM.2107-7520781/1100:IBM.2107-75ABTV1/2100 Full Duplex - Metro Mirror IBM.2107-7520781/11 unknown Disabled Invalid
IBM.2107-7520781/1101:IBM.2107-75ABTV1/2101 Full Duplex - Metro Mirror IBM.2107-7520781/11 unknown Disabled Invalid
dscli>
You can add the -at tgt parameter to the rmpprc command to remove only the Metro Mirror
target volume; see Example 14-13. The commands are given to the DS HMC connected to
DS8000#2, which is the Metro Mirror target.
Example 14-14 shows the Metro Mirror source volume status after the rmpprc -at tgt
command completed and the result of a rmpprc -at src command. In this case, there are still
available paths. Therefore, the source status changed after the rmpprc -at tgt command
completed. If there are no available paths, the state of the Metro Mirror source volumes is
preserved. You have to give the command to the DS HMC connected to DS8000#1, which is
the Metro Mirror source.
Example 14-14 Metro Mirror source volume status after rmpprc with -at tgt and rmpprc with -at src
dscli> lspprc 1002
Date/Time: October 26, 2005 6:16:22 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
==================================================================================================
1002:2002 Full Duplex - Metro Mirror 10 unknown Disabled Invalid
142 IBM System Storage DS8000 Series: Copy Services in Open Environments
<< After rmpprc -at tgt command completes >>
If you do not remove the Metro Mirror pairs that are using the paths, the rmpprcpath
command fails; see Example 14-16.
Example 14-17 Removing paths while still having Metro Mirror pairs with force parameter
dscli> lspprc 1002
Date/Time: October 26, 2005 7:57:34 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
==================================================================================================
1002:2002 Full Duplex - Metro Mirror 10 unknown Disabled Invalid
dscli>
dscli> rmpprcpath -remotedev IBM.2107-75ABTV1 -quiet 10:20
Date/Time: October 26, 2005 7:58:22 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
CMUN03070E rmpprcpath: 10:20: Copy Services operation failure: pairs remain
dscli>
dscli> rmpprcpath -remotedev IBM.2107-75ABTV1 -quiet -force 10:20
Date/Time: October 26, 2005 8:06:22 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
CMUC00150I rmpprcpath: Remote Mirror and Copy path 10:20 successfully removed.
dscli> lspprc 1002
Date/Time: October 26, 2005 8:06:28 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
==================================================================================================
1002:2002 Full Duplex - Metro Mirror 10 unknown Disabled Invalid
144 IBM System Storage DS8000 Series: Copy Services in Open Environments
Date/Time: October 26, 2005 11:00:21 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
CMUC00157I pausepprc: Remote Mirror and Copy volume pair 1000:2000 relationship successfully paused.
CMUC00157I pausepprc: Remote Mirror and Copy volume pair 1001:2001 relationship successfully paused.
dscli>
dscli> lspprc 1000-1001 1100-1101
Date/Time: October 26, 2005 11:00:33 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
=======================================================================================================
1000:2000 Suspended Host Source Metro Mirror 10 unknown Disabled Invalid
1001:2001 Suspended Host Source Metro Mirror 10 unknown Disabled Invalid
1100:2100 Full Duplex - Metro Mirror 11 unknown Disabled Invalid
1101:2101 Full Duplex - Metro Mirror 11 unknown Disabled Invalid
Because the source DS8000 keeps track of all changed data on the source volume, you can
resume Metro Mirror operations at a later time. The resumepprc command resumes a Metro
Mirror relationship for a volume pair and restarts transferring data. You must specify the
remote mirror and copy type, such as Metro Mirror or Global Copy with the -type parameter;
see Example 14-19.
When resuming the Metro Mirror pairs, the state of the pairs is copy pending, and the way the
copy is done, data consistency at the target volumes is not kept. Therefore, you must take
specific action to keep data consistency at the recovery site while resuming Metro Mirror
pairs. Taking an initial FlashCopy at the recovery site is one of the ways to keep data
consistency.
Important: When adding paths with the mkpprcpath command you must specify all of the
paths that you want to use. This includes the existing paths. Otherwise, you lose those
definitions that were already there.
146 IBM System Storage DS8000 Series: Copy Services in Open Environments
To reduce the number of paths you can also use the mkpprcpath command. In Example 14-21
for each LSS pair (10:20 and 11:21), we remove one path (I0102:I0031) from the existing
three paths (I0143:I0010, I0213:I0140,I0102:I0031).
Planned outages
The planned outage procedures rely on two facts:
Metro Mirror source and target volumes are in a consistent and current state.
Both DS8000s are functional and reachable.
You can swap sites without any full copy operation by combining Metro Mirror initialization
modes.
Source Target
(A volumes) (B volumes)
Metro Mirror paths
LSS10 LSS20
1000 2000
Metro Mirror Pairs
1001 2001
1100 2100
Metro Mirror Pairs
1101 2101
DS8000#1 DS8000#2
-dev IBM.2107-7520781 -dev IBM.2107-75ABTV1
A planned site switch using the Metro Mirror Failover function involves the following steps. If
the site switch is because of an unplanned outage, then the procedure starts from step 4 on
page 150:
1. When the planned outage window is reached, the applications at the production site (A)
must be quiesced to cease all write I/O activity. This way, there are no more updates to the
source volumes. Depending on the host operating system, it may be necessary to
dismount the source volumes.
148 IBM System Storage DS8000 Series: Copy Services in Open Environments
2. Next make sure that all Metro Mirror pairs are in full duplex state. It is better to check on
both sites, on DS8000#1 and on DS8000#2. In order to do this, you must issue the lspprc
command to each DS HMC; see Example 14-22.
Example 14-22 Check Metro Mirror state at the production site and the recovery site
<< DS8000#1 >>
dscli> lspprc 1000-1001 1100-1101
Date/Time: October 27, 2005 7:09:43 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
==================================================================================================
1000:2000 Full Duplex - Metro Mirror 10 unknown Disabled Invalid
1001:2001 Full Duplex - Metro Mirror 10 unknown Disabled Invalid
1100:2100 Full Duplex - Metro Mirror 11 unknown Disabled Invalid
1101:2101 Full Duplex - Metro Mirror 11 unknown Disabled Invalid
3. You can give the commands freezepprc and unfreezepprc to ensure that no data can
possibly be transferred to the target volumes (B volumes); see Example 14-23. This
operation is optional because the application at the production site has been quiesced.
Therefore, no data is sent to the target volumes.
Note: In the case of an unplanned outage, before you issue the failoverpprc
command, you may consider disconnecting the physical links between the production
and the recovery sites. This ensures that no unexpected data transfer to the recovery
site will occur at all.
Note: The lspprc command shows Target in the State column only for a target volume.
In the case of a source volume, there is no indication.
150 IBM System Storage DS8000 Series: Copy Services in Open Environments
5. Create paths in the direction recovery site to production site (B → A); see Example 14-25.
You have to give the mkpprcpath command to the DS HMC connected to DS8000#2.
Although it is not strictly necessary to reverse the paths, we recommend that you do it so
that you have a well-defined situation at the end of the procedure. Additionally, you will
need the paths to transfer the updates back to the production site.
A switchback using the Metro Mirror Failback function involves the following steps:
1. Using the lspprcpath command, verify that the paths from the recovery site to the
production site are available. Give this command to the DS HMC connected to DS8000#2;
see Example 14-26. You can check whether you have paths between the correct DS8000s
with the -fullid parameter.
2. If you did not reverse the paths before (in step 5 on page 151 of the previous procedure),
now you must establish paths from the recovery to the production site, before running the
failbackpprc command.
3. Run the failbackpprc command from the recovery site to the production site. You have to
give this command to the DS HMC connected to DS8000#2. The failbackpprc command
will copy all the modified tracks from the B volumes to the A volumes. See Example 14-27.
You must issue the failbackpprc command according to the roles the volumes will have
after the command is completed. In our example, you must specify the B volumes as the
source volumes and the A volumes as the target volumes. After the failbackpprc
command is successfully executed, the B volumes become source volumes in copy
pending state, and the A volumes become target volumes in target copy pending state.
Note: When issuing the failbackpprc command, if you specify the A volume as the
source and the B volume as the target, and you give the command to the DS HMC
connected to the DS8000#1, then data will be copied from A to B.
152 IBM System Storage DS8000 Series: Copy Services in Open Environments
dscli>
dscli> lspprc 2000-2001 2100-2101
Date/Time: October 28, 2005 12:22:25 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-75ABTV1
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
=====================================================================================================
2000:1000 Suspended Host Source Metro Mirror 20 unknown Disabled Invalid
2001:1001 Suspended Host Source Metro Mirror 20 unknown Disabled Invalid
2100:1100 Suspended Host Source Metro Mirror 21 unknown Disabled Invalid
2101:1101 Suspended Host Source Metro Mirror 21 unknown Disabled Invalid
dscli>
dscli> failbackpprc -remotedev IBM.2107-7520781 -type mmir 2000-2001:1000-1001 2100-2101:1100-1101
Date/Time: October 28, 2005 12:23:01 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-75ABTV1
CMUC00197I failbackpprc: Remote Mirror and Copy pair 2000:1000 successfully failed back.
CMUC00197I failbackpprc: Remote Mirror and Copy pair 2001:1001 successfully failed back.
CMUC00197I failbackpprc: Remote Mirror and Copy pair 2100:1100 successfully failed back.
CMUC00197I failbackpprc: Remote Mirror and Copy pair 2101:1101 successfully failed back.
dscli>
dscli> lspprc 2000-2001 2100-2101
Date/Time: October 28, 2005 12:23:05 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-75ABTV1
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
===================================================================================================
2000:1000 Copy Pending - Metro Mirror 20 unknown Disabled Invalid
2001:1001 Copy Pending - Metro Mirror 20 unknown Disabled Invalid
2100:1100 Copy Pending - Metro Mirror 21 unknown Disabled Invalid
2101:1101 Copy Pending - Metro Mirror 21 unknown Disabled Invalid
<< After performing varyoffvg A volumes from the AIX servers at the production site >>
2. Before returning to normal operation, the applications, still updating B volumes at the
recovery site, must be quiesced to cease all write I/O from updating the B volumes.
Depending on the host operating system, it may be necessary to dismount the B volumes.
154 IBM System Storage DS8000 Series: Copy Services in Open Environments
3. You should now execute one more failoverpprc command. At this time, you must specify
the A volumes as the source volumes and the B volumes as the target volumes. You must
give this command to the DS HMC connected to the DS8000#1.
This operation changes the state of the A volumes from target full duplex to (source)
suspended. The state of the B volumes is preserved; see Example 14-30.
4. Define paths in the direction of production site to recovery site (A → B); see
Example 14-31. You must create paths if you executed the freezepprc command in the
optional step 3 on page 151 of the production to recovery site switchover procedure
(freezepprc removed the paths).
6. Wait until the Metro Mirror pairs are synchronized. Normally, this operation does not take
much time because no data transfer is necessary. After the Metro Mirror pairs are
synchronized, the state of the sources volumes (A) becomes full duplex and the state of
the target volumes (B) becomes target full duplex; see Example 14-33.
Example 14-33 After the Metro Mirror pairs have been synchronized
<< DS8000#1 >>
dscli> lspprc 1000-1001 1100-1101
Date/Time: October 28, 2005 10:35:01 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
==================================================================================================
1000:2000 Full Duplex - Metro Mirror 10 unknown Disabled Invalid
1001:2001 Full Duplex - Metro Mirror 10 unknown Disabled Invalid
1100:2100 Full Duplex - Metro Mirror 11 unknown Disabled Invalid
1101:2101 Full Duplex - Metro Mirror 11 unknown Disabled Invalid
156 IBM System Storage DS8000 Series: Copy Services in Open Environments
6. failoverpprc A to B (to DS8000#1).
7. mkpprcpath A to B.
8. failbackpprc A to B (to DS8000#1).
9. Wait for volume pairs synchronization.
10.Start applications at the production site (A).
11.Eventually rmpprcpath B to A.
You can specify the Consistency Group option when you define the Metro Mirror paths that
communicate source and target LSSs, or when you change the default Consistency Group
setting on each LSS by means of the chlss command —by default, this option is disabled.
With the chlss command you can also change the default Consistency Group timeout value,
by means of the -extlongbusy parameter.
Example 14-34 shows how to enable the Consistency Group option with the chlss command.
The setting of the option can be verified with the showlss command.
Example 14-34 Change the default Consistency Group setting with the chlss command
dscli> showlss 10
Date/Time: October 29, 2005 12:16:20 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID 10
Group 0
addrgrp 1
stgtype fb
confgvols 4
subsys 0xFF10
pprcconsistgrp Disabled
xtndlbztimout 120 secs
dscli> showlss 11
Date/Time: October 29, 2005 12:16:24 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID 11
Group 1
addrgrp 1
stgtype fb
confgvols 4
subsys 0xFF11
pprcconsistgrp Disabled
xtndlbztimout 120 secs
dscli>
dscli> chlss -pprcconsistgrp enable 10-11
Date/Time: October 29, 2005 12:16:59 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
CMUC00029I chlss: LSS 11 successfully modified.
CMUC00029I chlss: LSS 10 successfully modified.
dscli>
dscli> showlss 10
Date/Time: October 29, 2005 12:17:06 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID 10
Group 0
addrgrp 1
stgtype fb
When the DS8000 detects a condition where it cannot update a Metro Mirror target volume, it
notifies this condition with and SNMP alert. At that moment, if an automation procedure is in
place, the SNMP alert will trigger the automation procedure which will then execute a
freezepprc command as the one showed in Example 14-35.
With the freezepprc command, the DS8000 holds the I/O activity to the addressed LSSs by
putting the source volumes in a queue full state for a time period. Example 14-36 shows, for
an AIX environment, what the iostat command reports during this time interval.
Example 14-36 AIX iostat command output report during the queue full condition
# lsvpcfg
vpath8 (Avail pv ) 75207811000 = hdisk18 (Avail ) hdisk22 (Avail )
vpath9 (Avail pv ) 75207811001 = hdisk19 (Avail ) hdisk23 (Avail )
vpath10 (Avail pv ) 75207811100 = hdisk20 (Avail ) hdisk24 (Avail )
vpath11 (Avail pv ) 75207811101 = hdisk21 (Avail ) hdisk25 (Avail )
158 IBM System Storage DS8000 Series: Copy Services in Open Environments
hdisk22 100.0 0.0 0.0 0 0
hdisk23 0.0 0.0 0.0 0 0
hdisk24 0.0 0.0 0.0 0 0
Note: in addition to holding (freezing) the I/O activity, the freezepprc command also
removes the paths between the affected LSSs.
After the freezepprc for all related LSSs completes, you have consistent data at the recovery
site. Therefore, at this moment the automation procedure can execute the unfreezepprc
command to release (thaw) the I/O that was on hold (frozen) on the affected LSSs.
Example 14-37 shows the unfreezepprc command that thaws the frozen I/O queue on the
LSS pairs 10:20 and 11:21.
In a situation where the data could not be replicated because of a links failure circumstance,
that is the production site kept running, then Metro Mirror processing can resume after the
links are recovered. Still, if the automation was triggered and the freezepprc was executed,
then the Metro Mirror paths have to be defined again. This is because the freezepprc
command removes the paths between the affected LSSs.
Then, after the paths are re-established you can execute a resumepprc command to
re-synchronize the Metro Mirror pairs. Example 14-38 shows this scenario.
Note: while doing the re-synchronization, the volume pairs will be in a copy pending state.
During this time period there will be no data consistency in the target volumes. Therefore,
you may want to take specific action to keep data consistency at the recovery site while
resuming the Metro Mirror pairs. First doing a FlashCopy at the recovery site is one way to
accomplish this.
You will see the Paths panel; see Figure 14-5. This panel shows that for the selected LSS 10
at the moment there are no paths defined.
In the Paths panel, from the Select Action pull-down, select Create to proceed with the first
step of the creation wizard.
160 IBM System Storage DS8000 Series: Copy Services in Open Environments
The creation wizard then displays the Select source LSS panel. Here you select the source
LSS; see Figure 14-6.
When the creation wizard displays the Select target LSS panel, you select from the pull-down
lists the storage complex, then the storage unit, then the storage image, and finally the target
LSS; see Figure 14-7.
When the creation wizard displays the Select source I/O ports panel, select using the check
boxes at least one I/O port (two is better) to use for Metro Mirror replication; see Figure 14-8
on page 162.
When the creation wizard displays the Select target I/Os ports panel, for each I/O port
selected during the third step, select from its related pull-down list the target I/O port; see
Figure 14-9.
162 IBM System Storage DS8000 Series: Copy Services in Open Environments
When the creation wizard displays the Select path options panel, you can use the check box
to select the option Define as Consistency Group; see Figure 14-10.
See 12.4, “Consistency Group function” on page 116, for a detailed discussion of this option.
Click Next to proceed with the sixth and last step of this wizard.
Figure 14-11 shows the Verification panel. Here you check all the components of your path
configuration and, if necessary, click Back to correct any of them or click Finish to validate
the configuration and end the wizard.
We will see now an example of how to add a path to an LSS pair which already has a path
defined. From the DS Storage Manager GUI you follow these steps:
1. Select Real-time manager.
2. Select Copy services.
3. Click Paths.
4. Select the source Storage Complex, Storage Unit, Storage Image, and LSS.
164 IBM System Storage DS8000 Series: Copy Services in Open Environments
You will see the Paths panel; see Figure 14-13. This panel shows one existing path for the
LSS pair 11:10.
In the Paths panel, from the Select Action pull-down, select Create to proceed with the first
step of the creation wizard.
The creation wizard then displays the Select source LSS panel. Here we select the source
LSS 11; see Figure 14-14.
When the creation wizard displays the Select source I/O ports panel, select using the check
boxes the port on the source for the additional path that you are defining; see Figure 14-16.
166 IBM System Storage DS8000 Series: Copy Services in Open Environments
When the creation wizard displays the Select target I/Os ports panel, you select from the
pull-down list the target I/O port; see Figure 14-17.
When the creation wizard displays the Select path options panel, you can use the check box
to select the option Define as Consistency Group; see Figure 14-18.
See 12.4, “Consistency Group function” on page 116, for a detailed discussion of this option.
Click Next to proceed with the sixth and last step of this wizard.
After you finish you are again presented the Paths panel, and now you will see the new path
recently added; see Figure 14-20.
Note: In our example, the defined paths are in one direction. If you want to mirror from the
target back to the source you will have to establish paths in the other direction.
168 IBM System Storage DS8000 Series: Copy Services in Open Environments
3. Click Paths.
4. Select the source Storage Complex, Storage Unit, Storage Image, and LSS.
You will see the Paths panel; see Figure 14-21. This panel shows a list of the existing paths
for the selected LSS 11. In this panel we check the box for the path we want to work with and
from the Select Action pull-down list we select LSS copy options.
The LSS Copy Options panel is then displayed; see Figure 14-22. Here you can change the
options of the defined path: various timeout values, as well as the critical mode attribute. For
a discussion of these options see Chapter 12, “Metro Mirror options and configuration” on
page 113.
You will see the Paths panel. This panel shows the list of established paths for the selected
LSS 11 of our example. Here you check the box to select the paths you want to remove; see
Figure 14-23.
Then from the Select Action pull-down list you select Delete.
Note: Deleting paths reduces bandwidth. For a discussion on this topic see the
Chapter 13, “Metro Mirror performance and scalability” on page 129.
170 IBM System Storage DS8000 Series: Copy Services in Open Environments
After you click Continue the deletion is completed. Then you are presented again the Paths
panel where you will see the path is not anymore there; see Figure 14-25. If the path still
appears to be in the list, click Refresh to renew the screen.
Some things to remember when you are going to create the volume pairs, are the following:
All volumes, on the source and also on the target, must exist before starting to create the
Metro Mirror pairs.
Source and target volumes should be of the same size. The target volume can be bigger
than the source volume, but if this is the case, you cannot reverse the direction of the pair.
So different sizes of source and target volumes makes sense only for migration purposes.
Usually before you start creating the Metro Mirror pairs, you have to define the paths
between the source and target LSS, as explained in “Create paths” on page 160.
To start creating volume pairs, from the DS Storage Manager GUI you follow these steps:
1. Select Real-time manager.
2. Select Copy services.
3. Click Metro Mirror.
In the Metro Mirror panel, from the Select Action pull-down menu you select Create to
proceed with the first step of the creation wizard.
The Create Metro Mirror panel is displayed; see Figure 14-27. Here you select the volume
pairing method. If you select Automated volume pair assignment the system will later
automatically select target volumes of the same size as the source volume. If you select
Manual volume pair assignment you have to manually assign to each source volume a
target volume, one after the other. In our example we selected Automated volume pair
assignment.
Click Next to proceed with the second step of the creation wizard.
172 IBM System Storage DS8000 Series: Copy Services in Open Environments
The Select source volumes panel is displayed; see Figure 14-28. Here you can check the
boxes to select the source volumes of the Metro Mirror pairs. If you still have not established
paths between the source and target LSSs, you can do this now by clicking the Create Paths
button in this panel and proceeding as explained in 14.10.1, “Create paths” on page 160.
Click Next to proceed with the third step of the creation wizard.
The Select target volumes (Auto pairing) panel is displayed; see Figure 14-29. Here you can
select all target volumes from the available ones on the target LSS.
Figure 14-29 Create MM pair - Step 3: select target volumes (Auto pairing)
Click Next to proceed with the fourth step of the creation wizard.
Click Next to proceed with the fifth step of the creation wizard.
The Verification panel is displayed; see Figure 14-31. Here, you can verify the Metro Mirror
volume pairs just created. Click Finish to execute this configuration.
The Metro Mirror panel will be presented again; see Figure 14-32 on page 175. This time it
will show the list of volume pairs just created. You may see that the state of the volumes is
Copy pending. This indicates that the initial copy from the source to the target volumes is still
in progress. Click Refresh to see if the status has changed.
174 IBM System Storage DS8000 Series: Copy Services in Open Environments
After a period of time, depending on the size and number of volumes, all volumes will be
synchronized (in-sync). At that moment the State column of the Metro Mirror panel will show
Full duplex for the volumes, as illustrated in Figure 14-32.
Figure 14-32 Create MM pair - Volume pairs created and in full duplex state
Note: On very high workloads, a high number of established Metro Mirror volumes, or
sharing of ports for Metro Mirror and host attachments, you will possibly have higher
response times for your applications during this synchronization time.
To suspend volume pairs, from the DS Storage Manager GUI you follow these steps:
1. Select Real-time manager.
2. Select Copy services.
3. Click Metro Mirror.
7. The Suspend Metro Mirror panel is displayed; see Figure 14-34. Here you are prompted to
suspend the pair at either the source or the target. It is more usual to suspend at the
source, as shown in our example. Then click OK.
176 IBM System Storage DS8000 Series: Copy Services in Open Environments
8. Next the Metro Mirror panel is displayed, now showing the selected pairs in suspended
mode; see Figure 14-35.
To resume volume pairs, from the DS Storage Manager GUI you follow these steps:
1. Select Real-time manager.
2. Select Copy services.
3. Click Metro Mirror.
4. The Metro Mirror panel is displayed; see Figure 14-36 on page 178. Here you select the
source Storage Complex, Storage Unit, Storage Image, and Resource Type. Possible
Resource Type options are:
– Lss, from which you select the LSS number
– Host attachment, from which you select the host attachment name
– Volume Group, from which you select the Volume Group name
– Show All Volumes, from which you select All FB volumes or All CKD volumes
7. The Resume Metro Mirror panel is displayed; see Figure 14-37. Here you can now confirm
your choices. Click OK to proceed.
8. Next the Metro Mirror panel is displayed, now showing the selected volumes in full duplex
mode after the re-synchronization process is completed; see Figure 14-38. You may need
to refresh the screen to see when the volumes reach the full duplex state.
178 IBM System Storage DS8000 Series: Copy Services in Open Environments
14.10.8 Metro Mirror Failover
In the following example, we have two disk subsystems, a DS8000 (machine type 2107) and a
DS6000 (machine type 1750). Each disk subsystem has two volumes. The DS8000 is the
production unit. The DS6000 is the backup unit. See Table 14-2.
Now we will use these two machines to describe a sites switch scenario.
Note: If you try to access this volume on the same server where the previous source was
or still is, some operating systems have some problems with disks with the same serial
number or signature. In most cases, there is a procedure to do this; see Appendix A,
“Open systems specifics” on page 591.
In this example, we assume that the production site has failed. We decide to start production
processing using the backup servers at the recovery site. We need to start using the Metro
Mirror target volumes (B). These volumes must become source volumes since we will be
writing them, and these changes must eventually be mirrored back to the production site. The
procedure we follow is:
1. Select Real-time manager.
2. Select Copy services.
3. Click Metro Mirror.
4. Select the target Storage complex, Storage unit, Storage image, and Resource Type.
Possible Resource Type options are:
– LSS from which you select the LSS number.
– Host attachment from which you select the host attachment name.
– Volume Group from which you select the Volume Group name.
– Show All Volumes from which you select All FB volumes or All CKD volumes
5. Select the volume pairs that you want to fail over.
7. We now verify the volumes with which we are working. When ready, click OK to proceed,
as shown in Figure 14-40.
8. The failover operation now takes place. We can see the result in Figure 14-41. Note that
Figure 14-41 shows the status of the B volumes, which are suspended. The A volumes will
still be in full duplex. This is because the Metro Mirror Failover function does not care
about the state of the former source volumes (A). We are now able to start the servers at
the backup site, and they can access the B volumes on the backup site DS6000.
180 IBM System Storage DS8000 Series: Copy Services in Open Environments
14.10.9 Metro Mirror Failback
We run our production using servers at the recovery site, which access our backup disk
subsystem (B). We now get ready to switch processing back to the production site (A). The
first thing to do is mirror the changes back from the backup site (B) to the production site (A).
We do this using the Metro Mirror Failback function.
During the failback, changes are copied. You can click Refresh to see if the copy is
finished. When the copy is complete, the volumes at the backup site (B) will show as full
duplex source volumes, as depicted in Figure 14-44.
Note: If the failback fails, check whether your prior active server can still access the
new target volume, or has not reset the reserve on the volume, and if your Metro Mirror
path is established between these LSSs.
182 IBM System Storage DS8000 Series: Copy Services in Open Environments
If we display the status of the volumes at the production site (on serial 7503461), they will
appear as full duplex targets, as shown in Figure 14-45. The source machine is serial
13-00247 at the backup site.
Figure 14-45 Failback - View from the production site Storage Unit
8. At this point, we shut down the servers at the backup site. Once they are shutdown, we
know that no more changes are written to the volumes at the backup site. Now we repeat
the failover and failback process, but we do so from the production site storage unit (A).
In Figure 14-46, we select the volumes on the production storage unit and perform a
Failover.
When the final Failback is complete, the production site storage unit is now the source again
and the backup site storage unit is now the target again. This means the production site
storage unit (in our example, serial 75-03461) now looks like Figure 14-48. This shows serial
75-03461 as the source machine while serial 13-00247 is the target machine. If we instead
take the view from the backup site storage unit (in our example, serial 13-00247), it now looks
like it did in Figure 14-39 on page 180, prior to starting this whole exercise.
Figure 14-48 Production site storage unit at conclusion of disaster recovery exercise
Note: If the Failback fails, check whether your prior active server can still access the new
target volume, or has not reset the reserve on the volume, and if your Metro Mirror path is
established between these LSSs. You might need to use the Reset Reservation option
depicted in Figure 14-43.
184 IBM System Storage DS8000 Series: Copy Services in Open Environments
Part 5
2 Write acknowledge
Server
write
1 3
Primary Secondary
(source) (target)
4
LUN or Write to secondary LUN or
volume (non-synchronously) volume
With Global Copy, write operations complete on the source disk subsystem before they are
received by the target disk subsystem. This capability is designed to prevent the local
system’s performance from being affected by wait time from writes on the remote system.
Therefore, the source and target copies can be separated by any distance. Figure 15-1
illustrates how Global Copy operates:
1. The host server makes a write I/O to the source (local) DS8000. The write is staged
through cache and non-volatile storage (NVS).
2. The write returns as completed to the host server’s application.
3. At a later time, that is, in a non-synchronous manner, the source DS8000 sends the
necessary data so that the updates are reflected on the target (remote) volumes. The
updates are grouped in batches for efficient transmission.
4. The target DS8000 returns write complete to the source DS8000 when the updates are
secured in the target DS8000 cache and NVS. The source DS8000 then resets its Global
Copy change recording information.
188 IBM System Storage DS8000 Series: Copy Services in Open Environments
Note: The efficient extended distance mirroring technique of Global Copy is achieved with
sophisticated algorithms. For example, if changed data is in the cache, then Global Copy
sends only the changed sectors. There are also sophisticated queuing algorithms to
schedule the processing of updated the tracks for each volume and set the batches of
updates to be transmitted.
Simplex
Terminate Terminate
Establish Metro
Establish Global Mirror
Copy
Go to sync
Resync Terminate
Resync
Suspend Suspend
Suspened
Global
Global Copy
Copy is
is aa recommended
recommended solution
solution for
for remote
remote data
data copy,
copy, data
data
migration
migration and
and offsite
offsite backup
backup -- over
over continental
continental distances
distances without
without
impacting
impacting application
application performance.
performance.
ItIt can
can be
be used
used for
for application
application recovery
recovery implementations
implementations ifif application
application I/O
I/O
activity
activity can
can be
be quiesced
quiesced and
and non-zero
non-zero data
data loss
loss RPO
RPO is
is admissible.
admissible.
As summarized in Figure 15-3, Global Copy is a recommended solution when you want to
perform remote data copy, data migration, off-site backup, and transmission of inactive
database logs without impacting application performance, which is particularly relevant when
implemented over continental distances. Global Copy can also be used for application
recovery solutions based on periodic point-in-time copies of the data. This requires short
quiescings of the application’s I/O activity.
190 IBM System Storage DS8000 Series: Copy Services in Open Environments
16
When you establish a Global Copy pair, the following options are available:
No copy
This option does not copy data from the source to the target volume. This option assumes
that the volumes are already synchronized. The data synchronization is your responsibility
and the DS8000 does not check its synchronization.
Target read
This option allows host servers to read from the target volume. For a host server to read
the volume, the volume pair must be in full duplex state. This parameter applies to open
systems volumes. It does not apply to System z volumes.
In the open system’s file system environment, even if an application reads data from a file
system, a SCSI write command may be issued to the LUN in which the file system resides,
because some information, such as the last access time stamp may be updated by the file
system. In this case, even if you specify this option, the operation may fail.
Suspend after data synchronization
This option suspends the remote copy pairs after the data has been synchronized. You
can use this option with the Go-to-sync operation (the Go-to-sync operation is discussed
later in 16.1.5, “Converting a Global Copy pair to Metro Mirror” on page 193.
Wait option
This option delays the command response until the volumes are in one of the final states:
simplex, full duplex, suspended, target full duplex, target suspended. This parameter
cannot be used with -type gcp or -mode nocp, but it can be used with the Go-to-sync
operation.
Reset reserve on target
This option allows the establishment of a Global Copy relationship when the target volume
is reserved by another host. If this option is not specified and the target volume is
reserved, the command fails. This parameter can only be used with open system volumes.
192 IBM System Storage DS8000 Series: Copy Services in Open Environments
suspended. When resuming a Global Copy relationship, you can use the same options you
use to initially establish a Global Copy pair, except for the no copy option.
You can convert a Global Copy pair to a Metro Mirror pair using the DS CLI or DS GUI.
When terminating the Global Copy relationship to establish host access to the target volumes,
the first issue may cause loss of transactions. Since a file system’s or database’s consistency
depends on the correct ordering of write sequences, the second issue can cause inconsistent
volumes. Therefore, to use target volumes by the host systems, you must make them
point-in-time consistent:
The application must be quiesced and the volume pairs temporarily suspended. This is
necessary for ensuring consistency not only at the volume level, but also at the application
level.
The target volumes have to catch up to their source counterparts. Global Copy catch-up is
the name of the transition that occurs to a Global Copy pair when it goes from its normal
out-of-sync condition until it reaches a full sync condition. At the end of this transition, the
source and target volumes become fully synchronized.
You should now perform a FlashCopy of the target volumes onto tertiary volumes, and
then resume the Global Copy pairs. These tertiary volumes are then a consistent
point-in-time copy of the primary volumes.
Application running
2.Quiesce Application
source target
Restart Application
4.Restart Application
source target Now we have
consistent
Application running data
5.Take FlashCopy B to C. FlashCopy
Return to 1.
source target
Note: In a DB2® environment, you can use a function provided by DB2 to stop the
application write I/O, which is the set write suspend/resume command.
194 IBM System Storage DS8000 Series: Copy Services in Open Environments
pair and suspend the pair after the synchronization has completed. You can use the single
mkpprc -type mmir -suspend command to perform a Go-to-sync and suspend operation.
After this step, the Global Copy target (B) will be holding a consistent copy of the data.
4. Restart the application at the production site.
Now that you have a consistent copy of the data at the recovery site, while no updates to
the source (A) volumes are sent due to the suspend state, you can restart the application
at the production site.
5. Take a FlashCopy B to C.
In order to resume the Global Copy operation, we have to first perform a FlashCopy from
the Global Copy target volumes (B) to the FlashCopy target volumes (C) to keep
consistent data at the recovery site. Otherwise, further Global Copy data transmissions
will update the Global Copy targets (B), modifying the consistent copy that at the moment
we have at the recovery site.
You can use the Remote FlashCopy function provided by the DS8000 to issue the
FlashCopy command to the DS8000 at the recovery site via the Global Copy paths. (If you
use Remote FlashCopy, you do not need to set up a LAN connection to the remote
DS8000.) After taking the FlashCopy, you have a consistent copy in the FlashCopy target
volume (C). You can then resume the Global Copy normal operation.
FlashCopy
FlashCopy
DS CLI client
Recovery site DS8000
In Figure 16-2, four FlashCopy pairs are established at the recovery site. The FlashCopy
targets are Vol 1, Vol 2, Vol 3, and Vol 4. In this case, the DS CLI client issuing the FlashCopy
establish command is in the production site. When the DS CLI client is giving the FlashCopy
establish command, a disaster may occur. It may happen, although it is a very short window,
that the DS CLI client can issue the first and second FlashCopy establish commands before
As long as you issue the FlashCopy commands from the production site, the above situation
can occur when you use any FlashCopy establish command, such as the mkflash,
resyncflash, mkremoteflash, and resyncremoteflash commands.
In order to avoid or detect this situation, you can implement several procedures, which
include:
Have the DS CLI client at the recovery site issue the FlashCopy command to the DS8000
in the recovery site. We also recommend that you keep the DS CLI execution log at the
recovery site to determine what processes have been performed during a disaster.
If you prefer having the DS CLI client at the production site, keep the DS CLI execution log
at the recovery site to determine what processes have been performed during a disaster.
If you use the Incremental FlashCopy function, you can use the FlashCopy sequence
number. When you establish FlashCopy, you can add a certain FlashCopy sequence
number to the set of the FlashCopy pairs. This sequence number can be used as an
identifier for a FlashCopy relation or group of FlashCopy relations. When a disaster
occurs, you can find whether you have the same set of the FlashCopy or not from this
sequence number. However, for safer implementation, we also recommend that you keep
the DS CLI execution log at the recovery site to determine what processes have been
performed during a disaster.
You can get Global Copy acquiring either the Metro Mirror licensed function or the Global
Mirror licensed function. See 12.11, “Hardware requirements” on page 127, for more
information.
Interoperability
Global Copy pairs can only be established between disk subsystems of the same (or similar)
type and features. For example, a DS8000 can have a Global Copy pair with another DS8000,
a DS6000, an ESS 800, or an ESS 750. It cannot have a Global Copy pair with an RVA or an
ESS F20. Note that all disk subsystems must have the appropriate Global Copy feature
installed. If your DS8000 is being mirrored to an ESS disk subsystem, the ESS must have
PPRC Version 2 (which supports Fibre Channel links) with the appropriate licensed internal
code level (LIC).
196 IBM System Storage DS8000 Series: Copy Services in Open Environments
16.4 Global Copy connectivity - paths and links
Global Copy pairs are set up between volumes in LSSs, usually in different disk subsystems,
and these are normally in separate locations. A path (or group of paths) needs to be defined
between the source LSS and the target LSS. These logical paths are defined over physical
links between the disk subsystems.
When we define paths we do not specify any particular remote copy function — paths can be
used for Global Copy, Metro Mirror, or Global Mirror replication. Therefore, the discussion
presented in 12.5, “Metro Mirror paths and links” on page 121, is also valid for Global Copy.
We recommend its reading.
16.5 Bandwidth
To determine the bandwidth requirement for your Global Copy environment, you should
consider both the amount of data you need to transfer to the remote site and the available
window.
If using Global Copy for daily off-site backup with the technique mentioned in 16.2, “Creating
a consistent point-in-time copy” on page 193, you can estimate the needed window to be the
sum of the application quiesce time, plus the FlashCopy establishment time, plus the
application restart time. However, the FlashCopy background copy operation may have an
influence over the Global Copy operation. In addition, if you take a backup to tape of the
FlashCopy target volumes, this activity also influences the Global Copy activity and the
FlashCopy background copy performance. Therefore, for a safer estimation of your daily
off-site backup window, as a very conservative approach we recommend that you estimate
each time sequentially, such as Global Copy data transmission, application quiesce,
FlashCopy establishment, application restart, FlashCopy background copy operation, and
backing up FlashCopy target.
You can estimate the approximate amount of data to be sent to the target by calculating the
amount of write data to the source. Tools to assist with this are operating system dependent,
like iostat. See the IBM Redbook IThe IBM TotalStorage DS8000 Series: Implementation,
SG24-6786. Another method, but not so exact, is to monitor the traffic over the FC switches
using FC switch tools and other management tools, but remember that only writes are
mirrored by Global Copy. You can also get some feeling about the proportion of reads/writes
by issuing datapath query devstats on SDD-attached servers.
Finally, you can estimate how much network bandwidth is necessary for your Global Copy
solution by dividing the amount of data you need to transfer by the duration of the backup
window.
The FlashCopy relationship at the remote site may influence the performance of the Global
Copy operation. If you use the FlashCopy with the nocopy option at the recovery site, when
If not using the nocopy option, but resuming Global Copy before the FlashCopy background
copy finishes at the remote site, this background data movement may also influence the
Global Copy performance.
This may also help you when you use the freezepprc command. A freeze operation is
performed at the LSS level, causing all Global Copy volumes in that LSS to go into
suspended state with queue full condition and terminating all associated paths. If you assign
LSSs to each of your applications, you can control the impact of the queue full condition
caused by the freeze operation at an application level. If you put volumes used by several
applications into the same LSS, all those applications sharing the LSS, their Global Copy
volumes, will be affected by the queue full condition. You can issue one freezepprc
command to multiple LSSs. Therefore, you do not need to consolidate the number of LSSs
that your applications use.
In general, the freezepprc and unfreezepprc commands will be used in the Metro Mirror
environment or after converting from Global Copy mode to Metro Mirror mode. However, if
you set the Consistency Group option on an LSS, even when the volumes in the LSS are
Global Copy sources, the freeze operation triggered by the freezepprc command, or a failure
of a target update, causes the queue full condition. Therefore, you should pay attention to the
LSS design if you use the Consistency Group option.
16.7 Distance
The non-synchronous characteristics of Global Copy, combined with the great throughput
characteristic of its efficient track mirroring technique, make Global Copy well suited for
remote copy solutions at distances beyond the 300 km available with Metro Mirror. Global
Copy achieves greater distances without having to incur in the distance latency penalties that
synchronous write I/O methods exhibit.
The maximum distance for a direct Fibre Channel connection is 10 km. If you want to use
Global Copy over longer distances, the following connectivity technologies can be used to
extend this distance:
Fibre Channel switches
Channel Extenders over Wide Area Network (WAN) lines
Dense Wave Division Multiplexors (DWDM) on dark fibre
198 IBM System Storage DS8000 Series: Copy Services in Open Environments
Channel extender
Channel extender vendors connect DS8000 systems with a variety of Wide Area Network
(WAN) connections, including Fibre Channel, Ethernet/IP, ATM-OC3, and T1/T3.
When you use channel extender products with Global Copy, the channel extender vendor
determines the maximum distance supported between the source and target DS8000. The
channel extender vendor should be contacted for its distance capability, line quality
requirements, and WAN attachment capabilities.
A complete and current list of Global Copy supported environments, configurations, networks,
and products is available in the DS8000 Interoperability Matrix at:
http://www-1.ibm.com/servers/storage/disk/ds8000/interop.html
You should also consult the channel extender vendor about hardware and software
prerequisites if you are using their products in a DS8000 Global Copy configuration.
Evaluation, qualification, approval, and support of Global Copy configurations using channel
extender products is the sole responsibility of the channel extender vendor.
A simple way to envision DWDM is to consider that, at the sending end, multiple fibre optic
input channels (such as ESCON, Fibre Channel, FICON, or Gbit Ethernet) are combined by
the DWDM into a single fibre optic cable. Each channel is encoded as light for a different
wavelength. Think of each individual channel as an individual color — the DWDM system is
transmitting a rainbow. At the receiving end, the DWDM fans out the different optical
channels. DWDM, by the very nature of its operation, provides the full bandwidth capability of
the individual channel. As the wavelength of light is, from a practical perspective. infinitely
divisible, DWDM technology is only limited by the sensitivity of its receptors, as the total
aggregate bandwidth possible.
A complete and current list of Global Copy supported environments, configurations, networks,
and products is available in the DS8000 Interoperability Matrix at:
http://www-1.ibm.com/servers/storage/disk/ds8000/interop.html
The DWDM vendor should be contacted regarding hardware and software prerequisites
when using their products in an DS8000 Global Copy configuration.
If you take backups of the FlashCopy target volumes onto tapes, you must ensure that the
tape resources are capable of handling these dump operations in between the point-in-time
checkpoints.
The information discussed in the present chapter can be complemented with the following:
Chapter 3, “DS Storage Manager (GUI)” on page 17
Chapter 4, “DS Command-Line Interface” on page 23
This chapter gives an overview of the DS CLI and the DS SM for Global Copy management.
DS CLI and DS SM are also covered in Part 2, “Interfaces” on page 15.
Also the following interfaces, not covered in this book, can be used:
DS Open Application Programming Interface (DS Open API)
z/OS interfaces: TSO, ICKDSF, and ANTRQST API
In order to use a z/OS interface to manage open systems LUNs, the DS8000 must have at
least one CKD volume. If you are interested in this possibility, you can refer to the IBM
Redbook The IBM TotalStorage DS8000 Series: Copy Services with IBM Eserver zSeries,
SG24-6787.
For information about the DS Open API, refer to the publication IBM System Storage DS
Open Application Programming Interface Reference, GC35-0516.
List available I/O ports that can be lsavailpprcport This information is shown
used to establish Global Copy during the process when a
paths. path is established.
202 IBM System Storage DS8000 Series: Copy Services in Open Environments
Task Command with Select option
DS CLI with DS SM
The DS CLI has the advantage that you can make scripts and use them for automation. The
DS Storage Manager GUI is a Web-based graphical user interface and is more intuitive than
the DS CLI.
When you establish or remove Global Copy paths and volume pairs, you must give the DS
CLI commands to the DS HMC that is connected to the source DS8000. Also, when checking
status information at the local and remote site, you must issue DS CLI list type commands,
such as the lspprc, to each DS8000, source, and target.
The DS CLI commands are documented in the publication IBM System Storage DS8000
Command-Line Interface User´s Guide, SC26-7916.
In this section we describe and give examples of the available DS CLI commands that you
can use for Global Copy setup and control.
Date/Time: November 15, 2005 5:40:09 PM EET IBM DSCLI Version: 5.1.0.204 DS:
IBM.2107-75ABTV1
Local Port Attached Port Type Switch ID Switch Port
===================================================
I0010 I0143 FCP NA NA
I0031 I0102 FCP NA NA
I0140 I0213 FCP NA NA
lspprcpath
The lspprcpath command shows you the established paths from an LSS. In Example 17-2
we list the paths from LSS 00.
Date/Time: November 15, 2005 5:40:56 PM EET IBM DSCLI Version: 5.1.0.204 DS:
IBM.2107-75ABTV1
Src Tgt State SS Port Attached Port Tgt WWNN
=========================================================
00 80 Success 8000 I0010 I0143 5005076303FFC1A5
00 80 Success 8000 I0031 I0102 5005076303FFC1A5
mkpprcpath
With the mkpprcpath command, you can establish paths between two LSSs. In Example 17-3
we establish a path from 2107-75ABTV1 LSS 00 to 2107-7520781 LSS 80.
Date/Time: November 15, 2005 5:52:34 PM EET IBM DSCLI Version: 5.1.0.204 DS:
IBM.2107-75ABTV1
CMUC00149I mkpprcpath: Remote Mirror and Copy path 00:80 successfully established.
rmpprcpath
The rmpprcpath command removes all paths between two LSSs; see Example 17-4.
Date/Time: November 15, 2005 5:51:11 PM EET IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-75ABTV1
CMUC00152W rmpprcpath: Are you sure you want to remove the Remote Mirror and Copy path 00:80:?
[y/n]:y
CMUC00150I rmpprcpath: Remote Mirror and Copy path 00:80 successfully removed.
204 IBM System Storage DS8000 Series: Copy Services in Open Environments
You can suppress confirmation message CMUC00152W by adding parameter -quiet in the
command.
failbackpprc
You can issue the failbackpprc command against any Global Copy source volume that is in a
suspended state. The command copies the required data from the source volume to the
target volume in order to resume mirroring; see Example 17-5.
This command is usually used after a failoverpprc command that has been issued to restart
mirroring either in the reverse direction (recovery site to production site) or original direction
(production site to recovery site). The command in Example 17-5 is given to a former target
volume 808A that was changed to (source) suspended by a preceding failoverpprc command.
After the command completes, 808A becomes the source volume and 000A the target
volume. Note that because the command reverses the copy direction, you first have to
establish paths in that direction.
Date/Time: November 16, 2005 10:23:04 AM EET IBM DSCLI Version: 5.1.0.204 DS:
IBM.2107-7520781
CMUC00197I failbackpprc: Remote Mirror and Copy pair 808A:000A successfully failed back.
failoverpprc
You can use the failoverpprc command to change a target device into (source) suspended
state while leaving the original source device in its current state; see Example 17-6.
Date/Time: November 16, 2005 9:33:01 AM EET IBM DSCLI Version: 5.1.0.204 DS:
IBM.2107-7520781
CMUC00196I failoverpprc: Remote Mirror and Copy pair 808A:000A successfully reversed.
lspprc
The lspprc command shows you which volumes are in a Global Copy relationship; see
Example 17-7.
Date/Time: November 16, 2005 11:02:55 AM EET IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-75ABTV1
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 000A:808A successfully
created.
freezepprc
With the freezpprc command (see Example 17-9), you can stop the write activity on all
Global Copy source and target volumes for a given source and target LSS. The command
creates a new Consistency Group. It places the source logical subsystem (LSS) in the long
busy state so that no I/Os can be directed to it. It sets the source volumes on the source LSS
in suspended state. Target volumes are left in their current state. The command also removes
paths between the source LSS and target LSS and sets the queue full condition for the
source volume. This causes the host to queue writes to the source volume until the queue full
condition is reset (with the unfreezepprc command). During the queue full condition, the
source volume reports long busy status.
The command in Example 17-9 freezes all I/O to volume pairs whose source volume is on
LSS 00 of IBM.2107-75ABTV1 and the target on LSS 80 of IBM.2107-7520781.
Date/Time: November 16, 2005 11:26:10 AM EET IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-75ABTV1
CMUC00161W freezepprc: Remote Mirror and Copy consistency group 00:80 successfully created.
pausepprc
The pausepprc command suspends a Global Copy pair; see Example 17-10.
Example 17-10 pausepprc command
dscli> pausepprc -dev ibm.2107-75abtv1 -remotedev ibm.2107-7520781 000a:808a
Date/Time: November 16, 2005 11:08:14 AM EET IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-75ABTV1
CMUC00157I pausepprc: Remote Mirror and Copy volume pair 000A:808A relationship successfully
paused.
resumepprc
With the resumepprc command you can bring a pair from suspended to copy pending mode,
as shown in Example 17-11.
Date/Time: November 16, 2005 11:14:43 AM EET IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-75ABTV1
CMUC00158I resumepprc: Remote Mirror and Copy volume pair 000A:808A relationship
successfully resumed. This message is being returned before the copy completes.
206 IBM System Storage DS8000 Series: Copy Services in Open Environments
rmpprc
The rmpprc command removes the Global Copy relationship from a volume pair and sets the
volumes in simplex status; see Example 17-12.
Date/Time: November 15, 2005 5:45:48 PM EET IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-75ABTV1
CMUC00160W rmpprc: Are you sure you want to delete the Remote Mirror and Copy volume pair
relationship 000A:808A:? [y/n]:y
CMUC00155I rmpprc: Remote Mirror and Copy volume pair 000A:808A relationship successfully
withdrawn.
You can suppress confirmation message CMUC00160W by adding parameter -quiet in the
command.
unfreezepprc
With the unfreezepprc command you can thaw an existing Consistency Group that was
created with the freeze command; see Example 17-13. It resumes host application I/O
activity for source volumes on the Consistency Group that was created by an earlier
freezepprc command. The unfreezepprc command is used to reset the queue full condition
for the source volume and release application I/O to the source volumes. All queued writes to
the source volume are written. The status of the volumes does not change; the source
volumes remain in the suspended state as set by the freezepprc command.
Example 17-13 unfreezepprc command
dscli> unfreezepprc -dev ibm.2107-75abtv1 -remotedev ibm.2107-7520781 00:80
Date/Time: November 16, 2005 1:11:30 PM EET IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-75ABTV1
CMUC00198I unfreezepprc: Remote Mirror and Copy pair 00:80 successfully thawed.
Note: Because the freezepprc command deletes the paths between the specified LSS
pair, you have to re-establish them to resume normal Global Copy operation.
For a detailed explanation of the DS SM Graphical User Interface, refer to Chapter 3, “DS
Storage Manager (GUI)” on page 17.
Note: The DS SM GUI supports, differently from the DS CLI, a multi-session environment,
but not all functions and options are supported by the GUI.
Then the Paths panel will display; see Figure 17-1. Here, to display existing paths, select the
storage complex, the storage unit, the storage image, and the source LSS for which you want
to display the paths information.
To create a new path, select Create from the Select Actions pull-down menu, and a process
will then guide you to establish a path.
Alternatively, if in the Paths panel you select one or more of the existing paths, then the Select
Actions pull-down menu will give you the following additional possible actions:
Delete, to delete the paths you have selected before
LSS copy options, to modify the copy options for an LSS (Figure 17-2)
208 IBM System Storage DS8000 Series: Copy Services in Open Environments
17.4.2 Metro Mirror panel
The Metro Mirror panel is the central point in the DS Storage Manager to establish new and
manage existing Global Copy pairs. You can get to the Metro Mirror panel as follows:
1. Select Real-time manager.
2. Select Copy services.
3. Click Metro Mirror.
Then the Metro Mirror panel will display; see Figure 17-3.
To create a new Global Copy pair, choose Create from the Select Action pull-down menu.
The DS Storage Manager will guide you through the process to create a new pair.
If you select an existing Global Copy pair (see Figure 17-4), then the Select Action pull-down
menu will show you additional actions that you can perform with the selected volume pair:
Delete: You can use this action to delete a Global Copy pair.
Suspend: You can use this action to suspend a Global Copy pair.
210 IBM System Storage DS8000 Series: Copy Services in Open Environments
18
You can estimate Global Copy application impact as that of the application when working with
Metro Mirror suspended volumes. For the DS8000, there is some more work to do with the
Global Copy volumes, as compared to the suspended volumes, because with Global Copy,
the changes have to be sent to the remote DS8000. But this is a negligible overhead for the
application, as compared with the typical synchronous overhead.
There will be no processor resources (CPU and memory) consumed by the Global Copy
volume pairs, as this is managed by the DS8000 subsystem.
If you take FlashCopy at the recovery site in your Global Copy implementation, you should
take into account the influence between Global Copy and the FlashCopy background copy;
refer to 16.5, “Bandwidth” on page 197.
18.2 Scalability
The DS8000 Global Copy environment can be scaled up or down as required. If new volumes
are added to the DS8000 that require mirroring, they can be dynamically added. If additional
Global Copy paths are required, they also can be dynamically added.
The eRCMF service offerings are designed to provide this functionality. Visit the IBM web site
and see the Services and Industry Solutions page for more information.
212 IBM System Storage DS8000 Series: Copy Services in Open Environments
19
LSS10 LSS20
1000 2000
1001 2001
1100 2100
1101 2101
DS8000#1 DS8000#2
-dev IBM.2107-7520781 -dev IBM.2107-75ABTV1
In our example we use different LSS and LUN numbers for the Global Copy source and target
elements, so that you can more clearly understand which one is being specified when going
through the reading of the example.
Note: In a real environment, and differently from our example, to simplify the management
of your Global Copy environment, we recommend that you maintain a symmetrical
configuration in terms of both physical and logical elements.
The procedure to set up the Global Copy environment is similar to that of a Metro Mirror
environment. One thing that is different is the type of relationship that is established between
the volumes. In this example we define Global Copy pairs.
214 IBM System Storage DS8000 Series: Copy Services in Open Environments
19.1.2 Set up of Global Copy configuration
Figure 19-2 shows the configuration we set up for this example. The configuration has four
Global Copy pairs that reside in two LSSs. Two paths are defined between each source and
target LSS.
1000 2000
Global Copy pairs
1001 2001
1100 2100
Global Copy pairs
1101 2101
DS8000#1 DS8000#2
-dev IBM.2107-7520781 -dev IBM.2107-75ABTV1
Example 19-1 Create Global Copy pairs and verify the result
dscli> mkpprc -remotedev IBM.2107-75ABTV1 -type gcp 1000-1001:2000-2001 1100-1101:2100-2101
Date/Time: November 2, 2005 10:53:22 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 1000:2000 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 1001:2001 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 1100:2100 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 1101:2101 successfully created.
dscli>
dscli> lspprc 1000-1001 1100-1101
Date/Time: November 2, 2005 10:57:30 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
You can check the status of Global Copy using the lspprc -l command. Out Of Sync Tracks
shows the remaining tracks to be sent to the target volume (the size of the logical track for the
FB volume for the DS8000 is 64 KB). You can use the lspprc -fullid command flag to
display the fully qualified DS8000 storage image ID in the command output; see
Example 19-2.
Example 19-2 lspprc -l and lspprc -fullid for Global Copy pairs
dscli> lspprc -l 1000-1001 1100-1101
Date/Time: November 2, 2005 10:57:37 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID State Reason Type Out Of Sync Tracks Tgt Read Src Cascade Tgt Cascade Date Suspended SourceLSS Timeout (secs)
Critical Mode First Pass Status
==========================================================================================================================================
1000:2000 Copy Pending - Global Copy 38379 Disabled Disabled invalid - 10 unknown
Disabled False
1001:2001 Copy Pending - Global Copy 38083 Disabled Disabled invalid - 10 unknown
Disabled False
1100:2100 Copy Pending - Global Copy 60840 Disabled Disabled invalid - 11 unknown
Disabled False
1101:2101 Copy Pending - Global Copy 60838 Disabled Disabled invalid - 11 unknown
Disabled False
dscli>
dscli> lspprc -fullid 1000-1001 1100-1101
Date/Time: November 2, 2005 11:06:40 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass
Status
==========================================================================================================================================
====
IBM.2107-7520781/1000:IBM.2107-75ABTV1/2000 Copy Pending - Global Copy IBM.2107-7520781/10 unknown Disabled True
IBM.2107-7520781/1001:IBM.2107-75ABTV1/2001 Copy Pending - Global Copy IBM.2107-7520781/10 unknown Disabled True
IBM.2107-7520781/1100:IBM.2107-75ABTV1/2100 Copy Pending - Global Copy IBM.2107-7520781/11 unknown Disabled True
IBM.2107-7520781/1101:IBM.2107-75ABTV1/2101 Copy Pending - Global Copy IBM.2107-7520781/11 unknown Disabled True
Unlike the Metro Mirror initial copy (first pass), the state of the Global Copy volumes still
shows Copy Pending even after the Out Of Sync Tracks become 0; see Example 19-3.
Example 19-3 List the Global Copy pairs status after Global Copy first pass completes
dscli> lspprc -l 1000-1001 1100-1101
Date/Time: November 2, 2005 10:58:43 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID State Reason Type Out Of Sync Tracks Tgt Read Src Cascade Tgt Cascade Date Suspended SourceLSS Timeout (secs)
Critical Mode First Pass Status
==========================================================================================================================================
216 IBM System Storage DS8000 Series: Copy Services in Open Environments
Copy Pending is the state of the Global Copy source. The target state is Target Copy
Pending. To see the target state in this example, you have to give the lspprc command to the
HMC connected to DS8000#2, which is the Global Copy target. See Example 19-4.
In general, the following steps should be done to clear the Global Copy environment.
1. Remove Global Copy pairs.
2. Remove logical paths.
Example 19-7 shows the Global Copy source volume status after the rmpprc -at tgt
command has completed and it also shows the result of a rmpprc -at src command. In this
case, there were still available paths. Therefore, the source volume state changed after the
rmpprc -at tgt command completed. If there were no available paths, the state of the Global
Copy source volumes would have been preserved. In Example 19-7, you must issue this
command to the HMC connected to DS8000#1, which is the Global Copy source.
Example 19-7 Global Copy source volume status after rmpprc with -at tgt and rmpprc with -at src
dscli> lspprc 1002
Date/Time: November 2, 2005 11:39:11 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
==================================================================================================
1002:2002 Copy Pending - Global Copy 10 unknown Disabled False
218 IBM System Storage DS8000 Series: Copy Services in Open Environments
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
==================================================================================================
1000:2000 Copy Pending - Global Copy 10 unknown Disabled True
1001:2001 Copy Pending - Global Copy 10 unknown Disabled True
1100:2100 Copy Pending - Global Copy 11 unknown Disabled True
1101:2101 Copy Pending - Global Copy 11 unknown Disabled True
dscli>
dscli> rmpprc -remotedev IBM.2107-75ABTV1 -quiet 1000-1001:2000-2001 1100-1101:2100-2101
Date/Time: November 3, 2005 1:29:51 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
CMUC00155I rmpprc: Remote Mirror and Copy volume pair 1000:2000 relationship successfully withdrawn.
CMUC00155I rmpprc: Remote Mirror and Copy volume pair 1001:2001 relationship successfully withdrawn.
CMUC00155I rmpprc: Remote Mirror and Copy volume pair 1100:2100 relationship successfully withdrawn.
CMUC00155I rmpprc: Remote Mirror and Copy volume pair 1101:2101 relationship successfully withdrawn.
dscli>
dscli> lspprc 1000-1001 1100-1101
Date/Time: November 3, 2005 1:29:54 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
CMUC00234I lspprc: No Remote Mirror and Copy found.
dscli>
dscli> rmpprcpath -quiet -remotedev IBM.2107-75ABTV1 10:20 11:21
Date/Time: November 3, 2005 1:31:09 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
CMUC00150I rmpprcpath: Remote Mirror and Copy path 10:20 successfully removed.
CMUC00150I rmpprcpath: Remote Mirror and Copy path 11:21 successfully removed.
If you do not remove the Global Copy pairs that are using the paths, the rmpprcpath
command fails; see Example 19-9.
Example 19-9 Remove paths without removing the Global Copy pairs
dscli> lspprc 1000-1001 1100-1101
Date/Time: November 3, 2005 1:38:19 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
==================================================================================================
1000:2000 Copy Pending - Global Copy 10 unknown Disabled True
1001:2001 Copy Pending - Global Copy 10 unknown Disabled True
1100:2100 Copy Pending - Global Copy 11 unknown Disabled True
1101:2101 Copy Pending - Global Copy 11 unknown Disabled True
dscli>
dscli> rmpprcpath -quiet -remotedev IBM.2107-75ABTV1 10:20 11:21
Date/Time: November 3, 2005 1:38:29 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
CMUN03070E rmpprcpath: 10:20: Copy Services operation failure: pairs remain
CMUN03070E rmpprcpath: 11:21: Copy Services operation failure: pairs remain
If you want to remove the paths still having the Global Copy pairs, you can use the -force
parameter; see Example 19-10. After the path has been removed, the Global Copy pair is still
in copy pending state until the source receives I/O from the servers. When I/O goes to the
Global Copy source, the source volume becomes suspended. If you set the Consistency
Group option for the LSS in which the volumes reside, I/Os from the servers are held with
queue full status for the specified timeout value.
Example 19-10 Remove paths still having Global Copy pairs - use -force parameter
dscli> lspprc 1000-1001 1100-1101
Date/Time: November 3, 2005 1:14:27 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
==================================================================================================
1000:2000 Copy Pending - Global Copy 10 unknown Disabled True
1001:2001 Copy Pending - Global Copy 10 unknown Disabled True
1100:2100 Copy Pending - Global Copy 11 unknown Disabled True
1101:2101 Copy Pending - Global Copy 11 unknown Disabled True
dscli> rmpprcpath -quiet -remotedev IBM.2107-75ABTV1 -force 10:20 11:21
Date/Time: November 3, 2005 1:17:46 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
CMUC00150I rmpprcpath: Remote Mirror and Copy path 10:20 successfully removed.
<< After I/O goes to the source volume(1000 and 1001) >>
220 IBM System Storage DS8000 Series: Copy Services in Open Environments
Because the source DS8000 keeps records of all changed data on the source volume, you
can resume Global Copy data transfer at a later time. The resumepprc command resumes a
Global Copy relationship for a volume pair and restarts transferring data. You must specify the
copy mode such as Metro Mirror or Global Copy with the -type parameter; see
Example 19-12.
Example 19-13 Change copy mode from Global Copy to Metro Mirror
dscli> lspprc 1000-1001 1100-1101
Date/Time: November 3, 2005 12:16:40 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
==================================================================================================
1000:2000 Copy Pending - Global Copy 10 unknown Disabled True
1001:2001 Copy Pending - Global Copy 10 unknown Disabled True
1100:2100 Copy Pending - Global Copy 11 unknown Disabled True
1101:2101 Copy Pending - Global Copy 11 unknown Disabled True
dscli>
dscli> mkpprc -remotedev IBM.2107-75ABTV1 -type mmir 1000-1001:2000-2001 1100-1101:2100-2101
Date/Time: November 3, 2005 12:30:10 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 1000:2000 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 1001:2001 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 1100:2100 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 1101:2101 successfully created.
dscli>
dscli> lspprc 1000-1001 1100-1101
Date/Time: November 3, 2005 12:30:14 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
==================================================================================================
You can use the -suspend parameter with the mkpprc -type mmir command. If you use this
parameter, the state of the pairs becomes suspended when the data synchronization is
completed; see Example 19-14. You can use this option for your off-site backup scenario with
the Global Copy function.
You can add the -wait parameter with the mkpprc command. With the -wait parameter, the
mkpprc -type mmir -suspend command does not return to the command prompt until the
pairs complete data synchronization and reach the suspended state; see Example 19-15.
Note: If you do not specify the -wait parameter with the mkpprc -type mmir -suspend com-
mand, the mkpprc command does not wait for the data synchronization. If you do not use
the -wait parameter, you must check the completion of the synchronization with the lspprc
command.
222 IBM System Storage DS8000 Series: Copy Services in Open Environments
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 1000:2000 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 1001:2001 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 1100:2100 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 1101:2101 successfully created.
1/4 pair 1001:2001 state: Suspended
2/4 pair 1000:2000 state: Suspended
3/4 pair 1101:2101 state: Suspended
4/4 pair 1100:2100 state: Suspended
dscli>
dscli> lspprc 1000-1001 1100-1101
Date/Time: November 3, 2005 12:48:56 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
=====================================================================================================
1000:2000 Suspended Host Source Metro Mirror 10 unknown Disabled Invalid
1001:2001 Suspended Host Source Metro Mirror 10 unknown Disabled Invalid
1100:2100 Suspended Host Source Metro Mirror 11 unknown Disabled Invalid
1101:2101 Suspended Host Source Metro Mirror 11 unknown Disabled Invalid
19.3.3 Change the copy mode from Metro Mirror to Global Copy
You cannot change the copy mode from the Metro Mirror to Global Copy directly. In order to
do this, you must use the pausepprc command to suspend the Metro Mirror pair first, and then
resume the pair in Global Copy mode with the resumepprc -type gcp command; see
Example 19-16.
Example 19-16 Change copy mode from Metro Mirror to Global Copy
dscli> lspprc 1000-1001 1100-1101
Date/Time: November 3, 2005 12:30:14 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
==================================================================================================
1000:2000 Full Duplex - Metro Mirror 10 unknown Disabled Invalid
1001:2001 Full Duplex - Metro Mirror 10 unknown Disabled Invalid
1100:2100 Full Duplex - Metro Mirror 11 unknown Disabled Invalid
1101:2101 Full Duplex - Metro Mirror 11 unknown Disabled Invalid
dscli>
dscli> mkpprc -remotedev IBM.2107-75ABTV1 -type gcp 1000-1001:2000-2001 1100-1101:2100-2101
Date/Time: November 3, 2005 12:33:55 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
CMUN03053E mkpprc: 1000:2000: Copy Services operation failure: invalid transition
CMUN03053E mkpprc: 1001:2001: Copy Services operation failure: invalid transition
CMUN03053E mkpprc: 1100:2100: Copy Services operation failure: invalid transition
CMUN03053E mkpprc: 1101:2101: Copy Services operation failure: invalid transition
dscli>
dscli> pausepprc -remotedev IBM.2107-75ABTV1 1000-1001:2000-2001 1100-1101:2100-2101
Date/Time: November 3, 2005 12:34:35 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
CMUC00157I pausepprc: Remote Mirror and Copy volume pair 1000:2000 relationship successfully paused.
CMUC00157I pausepprc: Remote Mirror and Copy volume pair 1001:2001 relationship successfully paused.
CMUC00157I pausepprc: Remote Mirror and Copy volume pair 1100:2100 relationship successfully paused.
CMUC00157I pausepprc: Remote Mirror and Copy volume pair 1101:2101 relationship successfully paused.
dscli>
dscli> lspprc 1000-1001 1100-1101
Date/Time: November 3, 2005 12:35:02 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
=====================================================================================================
1000:2000 Suspended Host Source Metro Mirror 10 unknown Disabled Invalid
1001:2001 Suspended Host Source Metro Mirror 10 unknown Disabled Invalid
1100:2100 Suspended Host Source Metro Mirror 11 unknown Disabled Invalid
1101:2101 Suspended Host Source Metro Mirror 11 unknown Disabled Invalid
dscli>
dscli> resumepprc -remotedev IBM.2107-75ABTV1 -type gcp 1000-1001:2000-2001 1100-1101:2100-2101
Date/Time: November 3, 2005 12:35:26 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
CMUC00158I resumepprc: Remote Mirror and Copy volume pair 1000:2000 relationship successfully resumed.
This message is being returned before the copy completes.
Figure 19-3 shows a diagram of the DS8000 Global Copy environment for this example. We
use the Remote Incremental FlashCopy function for the FlashCopy at the recovery site
(DS8000#2). We use 1000, 1001, 1100, and 1101 in DS8000#1 for Global Copy (GC)
sources; 2000, 2001, 2100, and 2101 in DS8000#2 for Global Copy targets and FlashCopy
(FC) sources; and 2002, 2003, 2102, and 2103 in the DS8000#2 for FlashCopy targets.
GC Target
and
GC Source FC source FC Target
Paths
LSS10 LSS20 FlashCopy
DS8000#1 DS8000#2
-dev IBM.2107-7520781 -dev IBM.2107-75ABTV1
Figure 19-3 The DS8000 environment for Global Copy offsite backup
224 IBM System Storage DS8000 Series: Copy Services in Open Environments
19.4.1 Initial setup for this environment
In order to set up the Global Copy and FlashCopy environment, we follow these steps:
1. Create paths, create pairs, and wait for the completion of the Global Copy initial copy.
2. Create the Incremental FlashCopy relationship at the recovery site and wait for the
completion of the FlashCopy background copy.
=====================================================================================================================
================================================
We then verify the FlashCopy background copy completion; see Example 19-19.
226 IBM System Storage DS8000 Series: Copy Services in Open Environments
19.4.2 Periodical backup operation
Depending on your backup requirement and application acceptance, you schedule the
following scenario periodically, such as daily. We show examples of the DS CLI commands to
control this procedure, based on the scenario illustrated in Figure 19-4.
Application running
2.Quiesce Application
source target
Restart Application
4.Restart Application
source target Now we have
consistent
Application running data
5.Take FlashCopy B to C. FlashCopy
Return to 1.
source target
For a detailed description of the preceding steps refer to 16.2.1, “Procedure to take a
consistent point-in-time copy” on page 194. Next we see how you can execute the procedure.
Step 1
In this step we are operate in the normal Global Copy mode. The lspprc and lsremoteflash
commands show the status in Example 19-20.
Step2
Depending on your application and platform, you should take the necessary actions to briefly
stop updates to the source volumes — quiesce application.
Step3
The DS CLI command examples for this step are shown in Example 19-21.
<< It may take time to show bellow messages depending on the amount of the Out Of Sync Tracks remained. >>
228 IBM System Storage DS8000 Series: Copy Services in Open Environments
1100:2100 Suspended Host Source Metro Mirror 0 Disabled Disabled invalid - 11
1101:2101 Suspended Host Source Metro Mirror 0 Disabled Disabled invalid - 11
dscli>
Step 4
Depending on your application and platform, you should take the necessary actions to restart
your application — resume I/O activity to the source volumes.
Step 5
The DS CLI command examples for this step are shown in Example 19-22 and
Example 19-23. We use the resyncremoteflash command to perform the Incremental
FlashCopy from the production site. We add the -seqnum parameter with the mkremoteflash
to add the FlashCopy sequence number to identify this FlashCopy set.
We specify the -type gcp parameter with the resumepprc command to restart the normal
Global Copy mode.
Note: To create and monitor Global Copy pairs, use the Metro Mirror panels.
230 IBM System Storage DS8000 Series: Copy Services in Open Environments
The Paths panel is displayed; see Figure 19-5. Because there are no listed paths, we know
that LSS 14 has no paths defined that use LSS 14 as the source LSS. In our example we
want to configure two paths from the DS8000 serial 7503461 LSS 14 to the DS8000 serial
7520781 LSS 18. From the Select Actions pull-down menu select Create.
We can now select the source LSS from the production DS8000; see Figure 19-6.
If your intended storage unit is not presented as an alternative then you must add the storage
complex that contains that storage unit. See Chapter 3, “DS Storage Manager (GUI)” on
page 17. First add the storage complex and then restart the process to create paths.
The Select source I/O ports panel is displayed. Here you can select the source I/O ports; see
Figure 19-8. Because we want to establish two paths, we select both I/O ports.
232 IBM System Storage DS8000 Series: Copy Services in Open Environments
The Select target I/O ports panel is displayed. Here you select the target ports. Because there
is only one target port available for each source port, the choices are made for us, as shown
in Figure 19-9.
The Select path options panel is displayed; see Figure 19-10. Here you can decide if you
want to create a Consistency Group. Because this is the setup of a Global Copy environment
we do not need a Consistency Group; therefore, we do not select this option.
After you finish you are again presented the Paths panel, and now you will see the existing
paths that you have just created; see Figure 19-12.
234 IBM System Storage DS8000 Series: Copy Services in Open Environments
The Metro Mirror panel will be displayed. See Figure 19-13. Here you select the source
Storage Complex, Storage Unit, Storage Image, and LSS. Then from the Select Action
pull-down menu select Create.
The Volume Pairing Method panel will be displayed; see Figure 19-14. Here we must select
either Automated volume pair assignment or Manual volume pair assignment. In our example
we select Automated volume pair assignment.
The Select target volumes (Auto pairing) panel will be displayed; see Figure 19-16. Here we
select the target Storage Complex, Storage Unit, Storage Image, and Resource Type. In our
example we chose the resource type LSS, so now we have to select the LSS number. After
this, we select the target volumes, which, because we chose automated pairing in the
previous step, will be automatically paired to source volumes.
236 IBM System Storage DS8000 Series: Copy Services in Open Environments
The Select copy options panel will be displayed; see Figure 19-17. Here we select Global
Copy and we also select Perform initial copy.
The Verification panel will be displayed; see Figure 19-18. Here we can verify the volume
pairs we configured and, if necessary, click Back to do corrections or click Finish to create
the Global Copy pairs.
238 IBM System Storage DS8000 Series: Copy Services in Open Environments
19.5.4 Convert to Metro Mirror (synchronous)
To have the volume pairs in a full duplex state, we need to synchronize them. In the Metro
Mirror panel we select the volumes we want to synchronize and then from the Select Actions
pull-down menu we choose Convert to synchronous. See Figure 19-21.
The Convert to synchronous confirmation panel will be displayed; see Figure 19-22. This
panel shows all the select volumes. Note that the Type column shows Metro Mirror. This is
because we are converting from non-synchronous Global Copy to synchronous Metro Mirror.
If everything is correct and you want to proceed with the synchronization, click OK to
continue.
In the Metro Mirror panel we select the volumes we want to suspend and then we choose
Suspend from the Select Action pull-down. We then get the choice to suspend the volumes
at the source or at the target. Normally we suspend at the source. We then click OK and the
volumes will suspend. Figure 19-24 shows the volumes already in suspended state.
Once the volumes are suspended we can restart the I/O at the application site. We can also
make a FlashCopy of the target volumes at the recovery site, and when finished we could
resume the pairs to re-establish the Global Copy replication again.
240 IBM System Storage DS8000 Series: Copy Services in Open Environments
Part 6
The information discussed in this chapter is complemented with the following IBM
publications and IBM Redbooks:
The IBM TotalStorage DS8000 Series: Implementation, SG24-6786
IBM System Storage DS8000 Command-Line Interface User´s Guide, SC26-7916
Still with the asynchronous data replication techniques, we must provide additional means to
ensure data consistency at the remote location. It even requires a solution that guarantees
data consistency not only within a single local-remote pair of storage disk subsystems but
also across multiple local and remote storage disk subsystems.
For a given pair of local and remote storage disk subsystems, a time stamp approach leads to
consistent data at the remote storage disk subsystem. Using this approach, and by sorting
the I/Os by their time stamps, the write I/Os can be applied at the remote disk subsystem in
the same sequence as they arrived at the local disk subsystem.
But when the application volumes are spread across multiple storage disk subsystems, this
time stamp concept alone is not sufficient to replicate data and provide data consistency at
the remote site. This additionally requires a Consistency Group concept.
In the rest of the present section we discuss how consistent data, that is dependent writes,
are managed either with a synchronous technique like Metro Mirror versus different
asynchronous techniques like Global Copy or Global Mirror.
Host
Server
1 4
Storage Disk Storage Disk
Subsystem 1 Subsystem 2
A
2C00
A1
A
Primary
Primary
2 Synchronous B1
Primary
Primary Replicate
source 3 target
local remote
Figure 20-1 Synchronous data replication
In synchronous data replication methods such as Metro Mirror, an application write always
goes through the following four steps; see Figure 20-1:
1. Write the data to the source storage disk subsystem cache. Note that this does not end
the I/O and does not present an I/O complete to the application.
244 IBM System Storage DS8000 Series: Copy Services in Open Environments
2. Replicate the data from the source storage disk subsystem cache to the target storage
disk subsystem cache.
3. Acknowledge to the source storage disk subsystem that data successfully arrived at the
target storage disk subsystem.
4. Present the successful I/O completion to the server, which is presented to the application
and concludes this I/O.
Now the next I/O that depends on the successful completion of the previous I/O can be
issued.
When you have dependent writes across multiple storage disk subsystems, synchronous
data replication alone does not guarantee that you can restart applications at the remote site
without doing some previous data recovery.
Consider a database environment that spreads across multiple storage disk subsystems at
the local or remote site; see Figure 20-2.
Database Database
4
Subsystem Subsystem
Switch over to remote site
source 1 target
2 Restart DB Subsystem 5
A1
A
Primary
Primary Synchronous B1
Primary
Log
Primary
1 Log' 6
Replicate
Recover A2
Storage Disk Storage Disk
Subsystem 3 Subsystem 4
2D00
A
A
2D00
A
3D00
A
A2
Primary
Primary
Primary
Primary
Primary
Synchronous
B2
Primary
Primary
Primary Replicate Primary
DB DB'
Figure 20-2 Synchronous data replication and unnecessary recovery after local site failure
A database update usually involves three dependent write I/Os to ensure data consistency,
even in the event of a failure during this process: see Figure 20-2:
1. Update intent to the logging files. Logging may happen to two logging files.
2. Update the data.
3. Indicate update complete to the logging files.
This does not happen with automated solutions such as GDPS (System z environments) and
eRCMF (open systems environments) that make use of the freeze capabilities of the IBM
System Storage DS8000. Had one of those automation softwares been there, after failing
over to the remote site, the database subsystem would have just restarted without the
necessity for a lengthy database recovery.
246 IBM System Storage DS8000 Series: Copy Services in Open Environments
Figure 20-3 illustrates a rolling disaster situation where we have an automation software that
makes use of the freeze capability of the DS8000.
Database Database
7
Subsystem Subsystem
Switch over to remote site
1 6 Recover A2
Restart
source Hold I/O target
Storage Disk 3 Storage Disk 8
Subsystem 1 Subsystem 2
8 8
A
2C00
4 3C00
A1
A
Primary
Primary Synchronous B1
Primary
Log
Primary 1 Log'
Redrive Replicate
I/O
Figure 20-3 Synchronous data replication, freeze, and restart without recovery required
This is different with an asynchronous data replication approach. We still require data
consistency at a distant site in order to just restart at the distant site when the local site
becomes unavailable.
An asynchronous remote copy approach is usually required when the distance between the
local site and the remote site is beyond an efficient distance for a synchronous remote copy
solution. Metro Mirror provides an efficient synchronous approach for up to 300 km, when
utilizing Fibre Channel links.
Host
Server
1
2 Storage Disk
Storage Disk
Subsystem 1 Subsystem 2
2C00
A
A1
A
Primary
Primary Asynchronous B1
Primary
Primary 3 Replicate
source 4 target
local remote
Figure 20-4 Asynchronous data replication
In an asynchronous data replication environment, an application write I/O goes through the
following steps; see Figure 20-4:
1. Write application data to the source storage disk subsystem cache.
248 IBM System Storage DS8000 Series: Copy Services in Open Environments
2. Present successful I/O completion to the host server. The application can then
immediately schedule the next I/O.
3. Replicate the data from the source storage disk subsystem cache to the target storage
disk subsystem cache.
4. Acknowledge to the source storage disk subsystem that data successfully arrived at the
target storage disk subsystem.
From the previous procedure, note how in an asynchronous type technique, the data
transmission and the I/O completion acknowledge are independent processes. This results in
virtually no application I/O impact, or at most to a minimal degree only. This also derives its
convenience when needing to replicate over long distances.
Host
Server
a b
1 2
Storage Disk
Subsystem 2
A
2C00
aPrimary
A
b Primary
Primary
Primary
Non synchronous
Replicate
Storage Disk 3 b
Subsystem 1
4
source target
local site
a in transit remote site
Figure 20-5 Global Copy sequence of data arrival not preserved at remote site
Global Copy by itself as a non synchronous data replication method does not provide data
consistency at the remote site.
In Figure 20-5 the sequence of data arrival at the local site is record b after record a. Due to
the way Global Copy cycles through its out-of-sync bit map, it may replicate record a to the
remote site after record b. This assumes that, for example, record a is written to a source disk
location that the Global Copy replication cycle just passed before. Global Copy may approach
another disk location that experienced just before an update with record b. Then record b gets
replicated to the remote site before the replication cycle starts from the beginning and finds
record a.
Database
subsystem
local site remote site
source target
1 3
Storage Disk Storage Disk
Subsystem 1 Subsystem 2
A
2C00
2C00 3C00
non synchronous
A
Primary
Primary
Primary
B
Primary
Log Log'
3 1
Replicate
Storage Disk 2 Storage Disk
Subsystem 3 Subsystem 4
2D00
A
A
2D00 non synchronous A
3D00
A
Primary
Primary
Primary
Primary
Primary B
Primary
Primary
Primary Primary
DB DB'
Replicate
Figure 20-6 Global Copy non synchronous data replication involving multiple disk subsystems
Notice that the numbers in Figure 20-6 indicate the transfer of data as well as the sequence of
its corresponding write I/Os. Record 3 may be replicated to storage disk subsystem 2 before
record 1 arrives to storage disk subsystem 2. Record 2 may never make it to storage disk
subsystem 4 due to the connectivity problem between storage disk subsystem 3 and storage
disk subsystem 4, which in turn may be the beginning of a rolling disaster. So, when using a
non-synchronous technique like Global Copy, consistent data cannot be guaranteed at the
remote site without any additional measures, functions, and procedures.
The challenge is to provide data consistency at the distant site at any time and independent of
the combination of storage disk subsystems involved. One solution is to combine Global Copy
with FlashCopy and create a three-copy solution. This requires a series of additional steps
that have to be organized and carried out:
1. At the local site, temporarily pause the application write I/Os on the source A volumes.
2. Wait for and make sure that the source A volumes and the target B volumes become
synchronized. Note the challenge to manage all volumes.
3. Using FlashCopy, create a point-in-time (PiT) copy of the B volumes at the remote site.
4. The FlashCopy targets, that is, the C volumes, will then hold a copy of the A volumes at
the time the application was paused or stopped. In this manner data consistency is kept at
the remote site.
5. Restart or resume the application write I/O activity to the A volumes.
250 IBM System Storage DS8000 Series: Copy Services in Open Environments
replication of data over long distances, without impacting the application I/O response time,
its operation would be transparent and autonomic from the users point of view, and most
importantly would provide a consistent copy of the data at the remote site at all times. All this
is what Global Mirror is about.
To accomplish the necessary activities with minimum impact on the application write I/O,
Global Mirror introduces a smart bitmap approach in the source storage disk subsystem. With
this, Global Mirror can resume the application I/O processing immediately after a very brief
serialization period for all involved source storage disk subsystems. This brief serialization
periodically occurs at the very beginning of a sequence of events that resemble the ones
outlined above. In the following chapters we explain in detail the characteristics of Global
Mirror, how it operates, and how can you manage it.
Summary
In summary, the following characteristics are what an asynchronous data replication
technique should provide:
Data replication to the remote site independent from application write I/O processing at the
local site. This derives in no impact, or at least only minimal impact, to the application local
write I/O response time.
Data consistency and dependent writes always maintained at the remote site.
Data currency at the remote site lags a little behind the local site. The remote site is
always less current than the local site. In peak write workloads this difference is going to
increase. Global Mirror does not throttle host write I/Os to eventually manage this
discrepancy.
The bandwidth requirement between local and remote sites does not have to be
configured for peak write workload. Link bandwidth utilization is improved over
synchronous solutions.
Tertiary copies are required at the remote site to preserve data consistency.
Data loss in disaster recovery situations is limited to the data in transit plus the data that
may still be in the queue at the local site waiting to be replicated to the recovery site.
Network
Task3
Task2
Client
Client
The server distributes the work to its clients. The server also coordinates all individual
feedback from the clients and decides based on this feedback further actions.
Looking at Figure 20-7, it is obvious that the communication paths between the server and all
its clients are key. Without communication paths between these four components the
functions will eventually come to a complete stop. Matters get more complicated when the
communication fails unexpectedly in the middle of information exchange between the server
and his clients or to some of his clients.
Usually a two-phase commit process helps to provide a consistent state for certain functions
and whether they have successfully completed at the client site. Once a function is
successfully completed and is acknowledged to the server, the server progresses to the next
function task. At the same time the server tries to do as much as possible in parallel to
minimize the hit on throughput due to serialization and checkpoints.
When certain activities are dependent on each other it is required that the server coordinate
these activities to ensure a proper sequence.
252 IBM System Storage DS8000 Series: Copy Services in Open Environments
The server and client can be replaced by terms such as master and subordinate; see
Figure 20-8. These terms are used later when discussing Global Mirror.
Network
FlashCopy2
FlashCopy2
Subordinate
Target
source
Figure 20-8 shows the basic Global Mirror structure. A master coordinates all efforts within a
Global Mirror environment. Once the master is started and manages a Global Mirror
environment, the master issues all related commands over inband communication to its
attached subordinates at the local site. This may include a subordinate within the master
itself. This communication between the master and an internal subordinate is transparent and
does not need any extra attention from the user. The subordinates use inband communication
to communicate with their related target storage disk subsystems at the remote site. The
master also receives all acknowledgements from his subordinates and their targets, and
coordinates and serializes all the activities in the session.
When the master and subordinate are in a single storage disk subsystem the subordinate is
internally managed by the master. With two or more storage disk subsystems at the local site,
which participate in a Global Mirror session, the subordinate is external and needs separate
attention when creating and managing a Global Mirror session or environment.
The following sections explain how Global Mirror works and how Global Mirror ensures
consistent data at any time at the remote site. First we go through the process of how to
create a Global Mirror environment. At the same time, this explanatio, gives us a first insight
on how Global Mirror works.
In order to understand how Global Mirror works, we start explaining first how a Global Mirror
environment, that is, a Global Mirror session, is created and started. This is a step-by-step
approach and helps to understand further the Global Mirror operational aspects.
Host
Write I/O
A
Primary
A
Primary
Primary
Primary
Primary
A
Primary
Primary
Primary
Host
Write I/O
A
Primary
A
Primary
Primary
Global Copy path
Primary
B
Primary
A
Primary
Primary
Primary
Note in Figure 20-10 that we establish Global Copy paths over an existing network. This
network may be based on an FCP transport technology or on an IP-based network.
Global Copy paths are logical connections that are defined over the physical links that
interconnect both sites. Note that all remote mirror and copy paths (that is Metro Mirror,
Global Copy, and Global Mirror paths) are similar and are based on FCP. The term Global
Copy path just denotes that the path is intended for Global Copy use.
254 IBM System Storage DS8000 Series: Copy Services in Open Environments
20.3.3 Create Global Copy relationships
Host
Write I/O
A
Primary
A
Primary
Primary
Primary
B
Primary
A
Primary Global Copy
target
source Primary
copy pending Primary
copy pending
O
O
S OOS: Out-of-sync bit map
Next, we create a Global Copy relationship between the source volume and the target
volume; see Figure 20-11. This step changes the target volume state from simplex (no
relationship) to target copy pending. This copy pending state applies to both volumes, source
copy pending and target copy pending. At the same time data starts to be copied from the
source volume to the target volume. After a first complete pass through the entire A volume,
Global Copy will constantly scan through the out-of-sync bit map. This bitmap indicates
changed data as it arrives from the applications to the source disk subsystem. Global Copy
replicates the data from the A volume to the B volume based on this out-of-sync bit map.
In the following paragraphs we refer to the source volume as the A volume and to the target
volume as the B volume for simplicity.
Global Copy does not immediately copy the data as it arrives to the A volume. Instead, this is
an asynchronous process. As soon as a track is changed by an application write I/O, it is
reflected in the out-of-sync bitmap as with all the other changed tracks. There can be several
concurrent replication processes that work through this bitmap, thus maximizing the utilization
of the high bandwidth Fibre Channel links.
This replication process keeps running until the Global Copy volume pair A-B is explicitly or
implicitly suspended or terminated. Note that a Global Mirror session command, for example,
to pause or to terminate a Global Mirror session, does not affect the Global Copy operation
between both volumes.
At this point data consistency does not yet exist at the remote site.
Host
Write I/O
FlashCopy
A
Primary
A
Primary
Primary
Primary
B
Primary
A
Primary Global Copy
target
source Primary
Primary
copy pending
copy pending C
tertiary
O
O SBM: Source Bit Map S T
S B B
TBM: Target Bit Map M M
The focus is now on the remote site. Figure 20-12 shows a FlashCopy relationship with a
Global Copy target volume as the FlashCopy source volume. Volume B is now both at the
same time a Global Copy target volume and a FlashCopy source volume. In the same disk
subsystem is the corresponding FlashCopy target volume.
Note that this FlashCopy relationship has certain attributes that are typical and required when
creating a Global Mirror session. These attributes are:
Inhibit target write: Protect the FlashCopy target volume from being modified by anyone
other than Global Mirror related actions.
Enable change recording: Apply changes only from the source volume to the target
volume that occurred to the source volume in between FlashCopy establish operations,
except for the first time when FlashCopy is initially established.
Make relationship persistent: Keep the FlashCopy relationship until explicitly or implicitly
terminated. This parameter is automatic due to the nocopy property.
Nocopy: Do not initiate background copy from source to target, but keep the set of
FlashCopy bitmaps required for tracking the source and target volumes. These bitmaps
are established at the first time when a FlashCopy relationship is created with the nocopy
attribute. Before a track in the source volume B is modified, between Consistency Group
creations, the track is copied to the target volume C to preserve the previous point-in-time
copy. This includes updates to the corresponding bitmaps to reflect the new location of the
track that belongs to the point-in-time copy. Note that each Global Copy write to its target
volume within the window of two adjacent Consistency Groups may cause FlashCopy I/O
operations.
256 IBM System Storage DS8000 Series: Copy Services in Open Environments
20.3.5 Define the Global Mirror session
Creating a Global Mirror session does not involve any volume within the local nor the remote
sites. Our focus is back onto the local site again; see Figure 20-13.
Host
FlashCopy
01 A
Primary
B
Primary
Global Copy
target
Primary
Primary
copy pending
C
tertiary
S T
B B
M M
Defining a Global Mirror session creates a kind of token, which is a number between 1 and
255. This number represents the Global Mirror session.
This session number is defined at the LSS level. Each LSS that has volumes that will be part
of the session needs a corresponding define session command. Currently only a single
session is allowed per storage disk subsystem. The architecture allows for more than one
session and will be exploited in the future.
Host
FlashCopy
01
A
A
Primary
Primary
B
Primary
A
Primary Global Copy
target
source Primary
copy pending Primary
copy pending C
tertiary
O
O S T
S B B
M M
This process adds source volumes to a list of volumes in the Global Mirror session. But at this
stage it still does not perform consistency group formation.
Note that Global Copy is replicating, on the Global Copy target volumes, the application
updates that arrive to the Global Copy source volumes. Initially the Global Copy source
volumes are placed in a join pending state. Once a Consistency Group is formed, the Global
Copy source volume will then be added to the session and will be placed in an in session
state.
Nothing happens to the C volume after its initial establishment in a Global Mirror session.
258 IBM System Storage DS8000 Series: Copy Services in Open Environments
20.3.7 Start the Global Mirror session
This is the last step and it starts the Global Mirror session. Upon this, Global Mirror starts to
form Consistency Groups at the remote site. As Figure 20-15 indicates, the focus here is on
the local site, with the start command issued to an LSS in the source storage disk subsystem.
With this start command you set the master storage disk subsystem and the master LSS.
From now on, session related commands have to go through this master LSS.
Host
FlashCopy
01
A
A
Primary
Primary
B
Primary
A
Primary Global Copy
target
source Primary
Primary
copy pending
copy pending C
tertiary
O
C
O S C T
R
S CR: Change recording bit map B R B
M M
This start command triggers events that involve all the volumes within the session. This
includes a very fast bitmap management at the local storage disk subsystem, issuing inband
FlashCopy commands from the local site to the remote site, and verifying that the
corresponding FlashCopy operations successfully finished. This all happens at the microcode
level of the related storage disk subsystems that are part of the session, fully transparently,
and in an autonomic fashion from the users’ perspective.
All C volumes that belong to the Global Mirror session comprise the Consistency Group. Now
let us see more details about the Consistency Group creation at the remote site.
The numbers in Figure 20-16 illustrate the sequence of the events involved in the creation of
a Consistency Group. This illustration provides a high-level view that is sufficient to
understand how this process works.
Serialize all
1 2 3 Perform
Global Copy Drain data from local to remote site
FlashCopy
source volumes
O
C A1 O B1
R source S target C1
tertiary
O
C
R A2 O B2
source S target C2
tertiary
Local Remote
Figure 20-16 Formation of consistent set of volumes at the remote site
Note that before step 1 and after step 3, Global Copy constantly scans through the
out-of-sync bitmaps and replicates data from A volumes to B volumes as described in 20.3.3,
“Create Global Copy relationships” on page 255.
When the creation of Consistency Group is triggered, the following steps occur in a serial
fashion:
1. Serialize all Global Copy source volumes. This imposes a brief hold on all incoming write
I/Os to all involved Global Copy source volumes. Once all source volumes are serialized,
the pause on the incoming write I/O is released and all further write I/Os are now noted in
the change recording bitmap. They are not replicated until step 3 is done, but application
write I/Os can immediately continue.
2. Drain includes the process to replicate all remaining data that is indicated in the
out-of-sync bitmap and still not replicated. Once all out-of-sync bitmaps are empty (note
that empty here does not mean in a literal sense) step 3 is triggered by the microcode from
the local site.
3. Now the B volumes contain all data as a quasi point-in-time copy, and are consistent due
to the serialization process in step 1 and the completed replication or drain process in step
2. Step 3 is now a FlashCopy that is triggered by the local system’s microcode as an
inband FlashCopy command to volume B, as FlashCopy source, and volume C, as
FlashCopy target volume. Note that this FlashCopy is a two-phase process: First, the
FlashCopy command to all involved FlashCopy pairs in the Global Mirror session. Then
the master collects the feedback and all incoming FlashCopy completion messages. When
all FlashCopy operations are successfully completed, then the master concludes that a
new Consistency Group has been successfully created.
260 IBM System Storage DS8000 Series: Copy Services in Open Environments
FlashCopy applies here only to changed data since the last FlashCopy operation. This is
because the start change recording property was set at the time when the FlashCopy
relationship was established. The FlashCopy relationship does not end due to the nocopy
property, which is also assigned at FlashCopy establish time; see 20.3.4, “Introduce
FlashCopy” on page 256. Note that the nocopy attribute results in that no physical
background copy is done for any tracks. Only bitmaps are maintained and updated.
Once step 3 is complete, a consistent set of volumes have been created at the remote. This
set of volumes, the C volumes, represents the Consistency Group.
At this very moment also, the B volumes and the C volumes are, only for this brief moment,
equal in their content. Immediately after the FlashCopy process is logically complete, the local
systems’s microcode is notified to continue with the Global Copy process. In order to replicate
the changes to the A volumes that occurred during the step 1 to step 3 window, the change
recording bitmap is mapped against the empty out-of-sync bitmap, and from now on, all
arriving write I/Os will end up again in the out-of-sync bitmap. From now on the conventional
Global Copy process, as outlined in 20.3.3, “Create Global Copy relationships” on page 255,
continues until the next Consistency Group creation process is started.
Maximum Maximum
coordination drain
time time
Serialize all
1 2 3 Perform
Global Copy Drain data from local to remote site
FlashCopy
souce volumes
CG interval
1 2 3 time
1 2 3
Global Mirror provides a set of three externalized parameters that can be used for tuning the
Consistency Group formation process, overriding the default values; see Figure 20-17:
Maximum coordination time: dictates, for the Global Copy source volumes that belong to
this Consistency Group, how long a host write I/O may be impacted due to coordination
and serialization activities. This time is measured in milli-seconds (ms). The default is
50 ms.
Maximum drain time: is the maximum time allowed for draining the out-of-sync bitmap
once the process to form a Consistency Group is started and step 1 successfully
completed. The maximum drain time is specified in units of seconds. The default is
30 seconds.
If the maximum drain time is exceeded, Global Mirror will fail to form the Consistency
Group and will evaluate the current throughput of the environment. If this indicates that
another drain failure is likely, Global Mirror will stay in Global Copy mode while regularly
There is no external parameter to limit the time for the FlashCopy operation. This is due to its
very fast bitmap manipulation process.
In the first step, the serialization step, when the Consistency Group spans over more than
one source disk subsystem, Global Mirror not only serializes all related Global Mirror source
volumes but also does the coordination with other storage disk subsystems.
Global Mirror utilizes a distributed approach as well as a two-phase commit technique for
activities between the master and its subordinate LSSs. The communication between the
local and the remote site is organized through the subordinate LSS. The subordinates
function here partly as a transient component for the Global Mirror activities, which are all
triggered and coordinated by the master. This distributed concept allows ways to provide a set
of data consistent volumes at the remote site independent of the number of involved storage
disk subsystems at the local or at the remote site.
262 IBM System Storage DS8000 Series: Copy Services in Open Environments
21
Global Mirror is intended for long distance data replication and, for this, it relies on the
network infrastructure and components used between the remote sites. This chapter does not
cover network-related topics. For information about these topics, refer to the IBM Redbook
IBM TotalStorage Business Continuity Solutions Guide, SG24-6547.
Dependent writes
If the start of one write operation is dependent upon the completion of a previous write, the
writes are dependent. Application examples for dependent writes are databases with their
associated logging files. For instance, the database logging file will be updated after a new
entry has been successfully written to a tablespace.
Compliance at the remote site, with the chronological order of dependent writes to source
volumes, is the basis to provide consistent data at the remote site through remote copy
operations.
Dependent writes is discussed in detail in 12.4.1, “Data consistency and dependent writes”
on page 117, in the Metro Mirror part.
Consistency
The consistency of data is ensured if the order of dependent writes to disks or disk groups is
maintained. With Copy Services solutions the data consistency at the remote site is important
for the usability of the data. Consistent data, for instance, gives the ability to perform a data
base restart rather than a data base recovery that could take hours or even days.
Data consistency across all target volumes spread across multiple storage disk subsystems
is essential for logical data integrity.
Data currency
This term describes the difference of time since the last data was written at the production
site, versus the time the same data was written to the remote site. It determines the amount of
data you have to recover at the remote site after a disaster. This is also called recovery point
objective or RPO. Only synchronous copy solutions such as Metro Mirror have a currency of
zero or RPO = zero. All asynchronous copy solutions have a data currency greater than zero.
With Global Mirror a data currency of a few seconds can be achieved, while data consistency
is always maintained by the Consistency Group process within Global Mirror.
For asynchronous replication, this means that the data is not replicated at the same time as
the local I/O happens, but the data is replicated with certain time lag. Examples of different
non-synchronous or asynchronous replication methods are:
Global Copy is a non-synchronous method that does not guarantee consistent data at the
remote site.
z/OS Global Mirror (formerly XRC) is an asynchronous replication method that guarantees
consistent data at the remote site.
Global Mirror is also an asynchronous replication method that provides consistent data at
the remote site.
Session
A Global Mirror session is a collection of volumes that are managed together when creating
consistent copies of data volumes. This set of volumes can reside in one or more LSSs and
one or more storage disk subsystems at the local site. Open systems volumes and z/OS
volumes can both be members of the same session.
264 IBM System Storage DS8000 Series: Copy Services in Open Environments
When you start or resume a session, the creation of Consistency Groups is performed, and
the master storage disk subsystem controls the session by communicating with the
subordinate storage disk subsystems.
There is also a session concept at the LSS level. But all LSS sessions are combined and
grouped together within a Global Mirror session.
Master
The master is a function inside a source storage disk subsystem that communicates with
subordinates in other storage disk subsystems, and controls the creation of Consistency
Groups while managing the Global Mirror session. The master is defined when the start
command for a session is issued to any LSS in a source storage disk subsystem. This
determines who becomes the master storage disk subsystem.
The master needs communication paths over Fibre Channel links to any one of the LSSs in
each subordinate disk storage subsystem.
Subordinate
The subordinate is a function inside a source storage disk subsystem that communicates with
the master and is controlled by the master. At least one of the LSSs of each subordinate
source storage disk subsystems needs Fibre Channel communication paths to the master.
This enables the communication between the master and the subordinate, and is required to
create Consistency Groups of volumes that spread more than one storage disk subsystem.
If all the volumes of a Global Mirror session reside in one source storage disk subsystem no
subordinate is required, because the master can communicate to all LSSs inside the source
storage disk subsystem.
Consistency Group
This is a group of volumes in one or more target storage disk subsystems whose data must
be kept consistent at the remote site.
Local Site
This is the site that contains the production servers. This and other publications use this term
interchangeably with the terms primary, production, or source site.
Remote Site
This is the site that contains the backup servers of a disaster recovery solution. This and
other publications use this term interchangeably with the terms secondary, backup, standby,
or target site.
Global Copy
01 A
Network
A Primary
B
Primary
Primary
Subordinate A
Primary
target Primary
Primary
source
copy pending
copy pending C
tertiary
paths
Local site Remote site(s)
01 A
A Primary
B
Primary
Master Primary
A
Primary
target Primary
Network Primary
source
copy pending
copy pending C
tertiary
Global Copy
Figure 21-1 Global Mirror basic configuration with master and subordinate disk subsystems
The order of commands to create a Global Mirror environment is not completely fixed and
allows for some variation. In order to be consistent with other sources and to not confuse the
user with a different sequence of commands, we recommend a meaningful order and suggest
the following steps to create a Global Mirror environment:
1. Define the paths between the local site and the remote site. In Figure 21-1, these are the
logical communication paths between corresponding LSSs at the local site and the remote
site, defined over Fibre Channel physical links that are configured over the network. Global
Copy source LSSs are represented by the A volumes and their corresponding Global
Copy target LSSs by the B volumes.
You may also define logical communication paths here between the master and any
subordinate storage disk subsystem that will be part of the Global Mirror session. Note
that these paths are defined between source storage disk subsystems at the local site.
With only a single source storage disk subsystem, you do not need to define paths to
connect internal LSSs within the source storage disk subsystem. The communication
between the master and the subordinates within a single source storage disk subsystem is
transparent and internally performed.
2. Once the communication paths are defined, start the Global Copy pairs that will be part of
a Global Mirror session. Global Copy has to be created with the mkpprc -type gcp
command. We recommend that you wait until the first initial copy is complete before you
continue to the next step. This avoids unnecessary FlashCopy background I/Os in the
following step.
266 IBM System Storage DS8000 Series: Copy Services in Open Environments
3. The next step is to create FlashCopy relationships between the B and C volumes. You can
use the mkremoteflash (abbreviated mkflash) command, with the following parameters:
-tgtinhibit, -record, and -nocp. The -persist parameter is automatically set when the
-record parameter is selected. For a discussion of the particular FlashCopy attributes that
are required for a Global Mirror FlashCopy see 20.3.4, “Introduce FlashCopy” on
page 256.
4. With external subordinates, that is, with more than one involved disk subsystem at the
local site, you need paths between the master LSS and any potential subordinate storage
disk subsystem at the local site. If you did not define these paths in the very first step, then
this is the time to create these paths before you continue with the next step.
5. Define a token that identifies the Global Mirror session. This is a session ID with a number
between 1 and 255. Define this session number to the master storage disk subsystem and
also to all potentially involved source LSSs that are going to be part of this Global Mirror
session and will contain Global Copy source volumes that will belong to the Global Mirror
session. All source LSSs include all LSSs in potential subordinate storage disk
subsystems that are going to be part of the Global Mirror session. For this step you can
use the mksession command.
6. The next step populates the session with the Global Copy source volumes. You should put
these Global Copy source volumes into the session after they complete their first pass for
initial copy. For this step you can use the chsession -action add -volume command.
7. Start the session. This command actually defines the master LSS. All further session
commands have to go through this LSS. You may specify with the start command the
Global Mirror tuning parameters such as maximum drain time, maximum coordination
time, and Consistency Group interval time. For this step you can use the mkgmir
command.
Note: You only add Global Copy source volumes to a Global Mirror session.
Volumes may be added to a session in any state, for example, simplex or pending. Volumes
that have not completed their initial copy phase stay in a join pending state until the first initial
copy is complete. If a volume in a session is suspended, it will cause Consistency Group
formation to fail.
We recommend that you add only Global Copy source volumes that have completed their
initial copy or first pass, although the microcode itself will stop volumes from joining the Global
Mirror session until the first pass is complete. Also, we recommend that you wait until the first
initial copy is complete before you create the FlashCopy relationship between the B and the C
volumes.
When you add a rather large number of volumes at once to an existing Global Mirror session,
then the available resources for Global Copy within the affected ranks may be utilized by the
initial copy pass. To minimize the impact to the production servers when adding many new
volumes, you may consider adding the new volumes to an existing Global Mirror session in
stages.
Suspending a Global Copy pair that belongs to an active Global Mirror session will impact the
formation of Consistency Groups. When you intend to remove Global Copy volumes from an
active Global Mirror session, follow these steps:
1. Remove the desired volumes from the Global Mirror session.
2. Withdraw the FlashCopy relationship between the B and C volumes.
3. Terminate the Global Copy pair to bring volume A and volume B into simplex mode.
Note: When you remove A volumes without pausing Global Mirror, you might see this
reflected as an error condition with the showmigr -metrics command, indicating that the
Consistency Group formation failed. However, this does not mean you have lost a
consistent copy at the remote site because Global Mirror does not take the FlashCopy (B
to C) for the failed Consistency Group data. This message indicates that just one
Consistency Group formation has failed, and Global Mirror will retry the sequence.
When you add a new LSS to an active session and this LSS belongs to a storage disk
subsystem that already has another LSS that belongs to this Global Mirror session, you may
add the new LSS to the session without stopping and starting the session again. This is true
for either the master or for a subordinate storage disk subsystem.
268 IBM System Storage DS8000 Series: Copy Services in Open Environments
When you give a start command after a pause command, as opposed to a resume, any value
for the Consistency Group interval time, maximum coordination interval time, or maximum
Consistency Group drain time in the start command is ignored. If these parameters have to
be altered, you have to give a resume command to the paused session with the parameters
specified. If you resume a paused session without specifying these parameters, they will be
set to their default values; see 20.4.2, “Consistency Group parameters” on page 261.
Important: When setting new values for the tuning parameters, be sure to check for errors
in Consistency Group formation and in draining the out-of-sync bitmaps. A few errors are
not significant and do not jeopardize the consistency of your Global Mirror. However, if
failures repeatedly occur, for example, no more consistency groups are formed, or the
percentage of successful Consistency Groups is unacceptable, or the frequency of
Consistency Groups is not meeting your requirements (Recovery Point Objective-RPO),
then the new values set for the tuning parameters need to be revised and changed.
If you need to change the topology of your Global Mirror session, you have to stop the session
and start the session again with the new topology structure information in the mkgmir
command.
Topology also refers to the list of storage disk subsystems that are subordinates. You define
control paths between the master and subordinate LSSs. One LSS per subordinate disk
subsystem is sufficient. When you define the control path you identify the source LSS on the
master disk subsystem. The target LSS in the path definition command points to a
corresponding subordinate disk subsystem. These LSSs go into the topology specification
that defines the communication paths between master and subordinate storage disk
subsystems. To change these values you must stop the Global Mirror process (rmgmir
command) and start it again with the new topology specifications (mkgmir command).
Once the volumes are removed from the session, you may explicitly terminate the
corresponding FlashCopy relationship that was tied to this source volume.
The termination of FlashCopy relationships, may be needed when you want to change the
FlashCopy targets within a Global Mirror configuration and choose, for example, another LSS
for the FlashCopy targets. This may be needed because you want to replace the FlashCopy
targets due to a skew in the load pattern in the remote storage disk subsystem. In this
situation you may pause the session before such activity, and then resume the session again
once the removal of the FlashCopy relationships is completed.
01 Network A
B
Primary
Primary
Primary
Primary
C
01 A
Primary
B
Primary
Primary
Network Primary
C
Figure 21-2 Define Global Mirror session to all potentially involved storage disk subsystems
270 IBM System Storage DS8000 Series: Copy Services in Open Environments
Figure 21-2 on page 270 shows a symmetrical configuration with a one-to-one mapping. You
have to define the corresponding session, with its number, to all potentially involved LSSs at
the local site. There is still no connection between both local storage disk subsystems.
Therefore we have to define the corresponding Global Mirror paths; see Figure 21-3.
Global Copy
01
Network A
A
Primary
Primary
B
Primary
Subordinate A
Primary
target Primary
Primary
source
copy pending
copy pending C
tertiary
01 A
A Primary
B
Primary
Master Primary
A
Primary
target Primary
Network Primary
source
copy pending
copy pending C
tertiary
Global Copy
Figure 21-3 Decide for master disk subsystem and start Global Mirror session
Through the start command mkgmir, you decide which LSS becomes the master LSS and
consequently which local storage disk subsystem becomes the master storage disk
subsystem. This master acts like a server in a client/server environment.
The required communication between the master storage disk subsystem and the
subordinate storage disk subsystems is inband, over the defined Global Mirror paths. This
communication is highly optimized, and minimizes any potential application write I/O impact
during the coordination phase to about a few milliseconds. For details see 20.4, “Consistency
Groups” on page 259.
01
A
A
Primary
Primary
B
Primary
Subordinate A
Primary
target Primary
Primary
source
copy pending
copy pending
Global Copy links C
tertiary
01
A
A
Primary
Primary
B
Primary
Subordinate A
Primary
target Primary
Primary
source
copy pending
copy pending
Global Copy links C
tertiary
GM links
01 A
A Network Primary
B
Primary
Master Primary
A
Primary
target Primary
Primary
source
copy pending Global Copy links
copy pending C
tertiary
Figure 21-4 Global Mirror paths over FCP links between source storage disk subsystems
Also shown in Figure 21-4 is a shared port on the master storage disk subsystem, and
dedicated ports at the subordinates. Not considering availability, from a communications and
traffic viewpoint only, a link would be sufficient for the traffic between the master and its
subordinates. For redundancy, we suggest configuring two links.
Note that when you configure links over a SAN network, the same FCP ports of the storage
disk subsystem may be used for the Global Mirror session communication, as well as for the
Global Copy communication, and for host connectivity. However, for performance reasons,
and to prevent host errors from disrupting your Global Mirror environment, it is often a good
idea to use separate FCP ports.
272 IBM System Storage DS8000 Series: Copy Services in Open Environments
The sample configuration shown in Figure 21-5 shows a mix of dedicated and shared FCP
ports. In this example an FCP port in the master storage disk subsystem is used as Global
Mirror link to the other two subordinate storage disk subsystems, and is also used as Global
Copy link to the target disk subsystem. Also, there are ports at the subordinate disk
subsystem that are used as Global Mirror session link as well as Global Copy link.
01
A
A
Primary
Primary
B
Primary
Subordinate A
Primary
target Primary
Primary
source
copy pending
copy pending
Global Copy links C
tertiary
01
A
A
Primary
Primary
B
Primary
Subordinate A
Primary
target Primary
Primary
source
copy pending
copy pending Global Copy links C
tertiary
GM links
01 A
A Network Primary
B
Primary
Master Primary
A
Primary
target Primary
Primary
source
copy pending Global Copy links
copy pending
C
tertiary
01
A
A Primary
B
Primary
Primary
Subordinate A
Primary
target Primary
Primary
source
copy pending
copy pending
Global Copy links C
tertiary
01
A
A
Primary
Primary
B
Primary
Subordinate A
Primary
target Primary
Primary
source
copy pending
copy pending Global Copy links C
tertiary
GM links
01 A
A Network Primary
B
Primary
Master Primary
A
Primary
target Primary
Primary
source
copy pending Global Copy links
copy pending
C
tertiary
Figure 21-6 Dedicated Global Mirror links and dedicated Global Copy links
274 IBM System Storage DS8000 Series: Copy Services in Open Environments
21.5.1 Normal Global Mirror operation
Figure 21-7 shows a simple configuration with Global Mirror active and running.
Server
FlashCopy
Global Copy A
A A
Primary
Primary
A
Primary
Primary
Primary
Primary
B
Primary
A
A
Primary
source
target
A
Primary
Primary
Primary
copy pending
copy pending C
Primary
tertiary
Writes from the server are replicated through Global Copy, and Consistency Groups are
created as tertiary copies. The B volumes are Global Copy target volumes, and they are also
FlashCopy source volumes. The C volumes are the FlashCopy target volumes. The
FlashCopy relationship is a particular relationship; see 20.3.4, “Introduce FlashCopy” on
page 256.
The goal is to swap to the remote site and restart the applications. This requires, first, to make
the set of consistent volumes at the remote site available for the application, before the
application can be restarted at the remote site.
Then, once the local site is back and operational again, we must return to the local site.
Before returning to the local site, we must apply to the source volumes the changes that the
application did to the target data while it was running at the remote site. After this, we should
do a quick swap back to the local site and restart the application.
FlashCopy
Global Copy A
A A
Primary
Primary
A
Primary
Primary
Primary
Primary
B
Primary
A
A
Primary target
A
Primary
source Primary
Primary
copy pending
copy pending C
Primary
tertiary
When the local storage disk subsystem fails, Global Mirror can no longer form Consistency
Groups. Depending on the state of the local storage disk subsystem, you may be able to
terminate the Global Mirror session. Usually this is not possible because the storage disk
subsystem may not respond any longer. Host application I/O may have failed and the
application ended. This goes usually along with messages or SNMP alerts that indicate the
problem. With automation in place, for example, eRCMF, these alert messages trigger the
initial recovery actions.
If the formation of a Consistency Group was in progress then, most probably, not all
FlashCopy relationships between the B an C volumes at the remote site will have reached the
corresponding point-in-time. Some FlashCopy pairs may have completed the FlashCopy
phase to form a new Consistency Group, and committed the changes already. Others may not
have completed yet, and are in the middle of forming their consistent copy, and remain in a
revertible state. And there is no master any longer to control and coordinate what may still be
going on. All of this imposes a closer look at the volumes at the remote site before we can
continue to work with them. There is more discussion of this in the following sections. First,
however, we bring the B volumes into a usable state using the failover command.
Server for
GM recovery
Failover B to A
FlashCopy
Global Copy A
A A
Primary
Primary
A
Primary
Primary
Primary
Primary
B
Primary
A
A
Primary
source
source
A
Primary
Primary
Primary
copy pending
suspended C
Primary
tertiary
276 IBM System Storage DS8000 Series: Copy Services in Open Environments
Because the source storage disk subsystem may no longer be usable, now the recovery
actions and processing occur at the remote site, using a server connected to the remote
storage disk subsystems for storage management console functions. See Figure 21-9 on
page 276.
You may use DS Storage Manager (DS SM), DS Command-Line Interface (DS CLI), and
eRCFM to execute the needed commands to the remote storage disk subsystems.
Doing a failover (Copy Services Failover function) on the Global Copy target volumes will turn
them into source volumes and suspend them immediately. This sets the stage for change
recording when application updates start changing the B volumes. Change recording in turn
allows you to re-synchronize the changes later from the B to the A volumes, before returning
to the local site to resume the application again at the local site. But at this stage the B
volumes do not contain consistent data — they are still useless. We just changed their state
from target copy pending to suspended. The A volumes state remains unchanged.
The key element when you run a Copy Services Failover is that the B volumes become the
new source volumes. This action just changes the state of the B volumes from target copy
pending to suspended. This action does not require communication with the other storage disk
subsystem at all, even though it is specified in the failoverpprc command. Once all the
failover commands are successfully executed, we can move on to the next step.
Server for
GM recovery
Check Consistency Group
FlashCopy
Global Copy A
A A
Primary
Primary
A
Primary
Primary
Primary
Primary
B
Primary
A
A
Primary
source
source
A
Primary
Primary
Primary
copy pending
suspended C
Primary
tertiary
Each FlashCopy pair needs a FlashCopy query to identify its state. If the local site is still
accessible and also the source storage disk subsystem is accessible, you may consider
carrying these activities at the production site using remote FlashCopy commands. Most
likely the source storage disk subsystem does not respond any longer. In this case you must
target the query directly to the recovery site storage disk subsystem.
The lsflash command reports the Revertible state as either Enable or Disable, which
indicates whether the state of the FlashCopy is revertible or non-revertible. A non-revertible
state means that a FlashCopy process has completed successfully and all changes are
committed. Global Mirror uses the two-phase FlashCopy establishment operation. This
operation allows the storage disk subsystem to prepare for a new FlashCopy relationship
without altering the existing FlashCopy relationship. You can either commit or revert this new
FlashCopy relationship with revertible state by using the revertflash and commitflash
commands. During the Consistency Group formation process, Global Mirror puts all
FlashCopy relationships in the revertible state, and after they are in the revertible state,
commits all FlashCopy relationships. With this operation, the situation in which some
FlashCopy operatons have not started while others have completed does not occur.
The Sequence number is an identifier that can be set at FlashCopy establish operations and it
is then associated with the FlashCopy relationship. Subsequent FlashCopy withdraw
operations can be directed to FlashCopy relationships with specific sequence numbers.
Global Mirror uses the sequence number to identify a particular Consistency Group. The
actual sequence number used by Global Mirror is the platform timer from the Global Mirror
master storage disk subsystem (in seconds resolution) at the point when the Global Mirror
source components have to be coordinated to form a Consistency Group. This is at a point
before the Consistency Group is transferred to the remote site. If your master storage disk
subsystem platform timer is set to the time of day, then the FlashCopy sequence for Global
Mirror approximates a time stamp for the Consistency Group.
The best situation is when all the FlashCopy pairs of a Global Mirror session are in the
non-revertible state and all their sequence numbers are equal. No further action is necessary,
and the set of C volumes is consistent, and the copy is good.
Figure 21-11 shows the consistency group creation process. The action required depends on
the state of the consistency group creation process when the failure occurred.
Drain consistency group and send FlashCopy committed Start next consistency group
to remote DS using Global Copy. once all revertible
Application writes for next Flashcopies have
consistency group are recorded in successfully completed
change recording bitmap Action required
Maximum drain time – eg.1 min
278 IBM System Storage DS8000 Series: Copy Services in Open Environments
Depending on when the failure occurs, there are some combinations of revertible states and
FlashCopy sequence numbers that need different corrective actions. Use Table 21-1 as a
guide. This is a decision table and reads in the following way: When column 2 and column 3
are true, then take the action in column 4. Column 5 contains additional comments. Do this for
each of the four cases. The cases are described in chronological order, starting from the left.
If you see a situation other than the above four situations, then the Global Mirror mechanism
has been corrupted.
The FlashCopy relationships that have not started are in a non-revertible state and all of them
have the same FlashCopy sequence number. Other FlashCopy relationships, which had
already started, are in a revertible state and all of them have the same FlashCopy sequence
number, but the number is different from the sequence number of the non-revertible
FlashCopy relationships.
All these indications suggest that you have to return the revertible FlashCopy relationships to
the previous Consistency Group using the revertflash command, and terminate the
FlashCopy relationships.
The revert action, which is invoked by the revertflash command, restores the Consistency
Group level between the source and target volumes to the prior state before the current
FlashCopy ran and resets the revertible state to Disable. The FlashCopy relationship is
preserved.
When the FlashCopy relationship is in a non-revertible state, the revert operation is not
possible. When you issue this command to FlashCopy pairs that are non-revertible, you are
going to see only an error message, but no action is performed.
To make the task easier, you may run the commitflash to all FlashCopy pairs. Non-revertible
FlashCopy relationships will just return an error message.
Usually all FlashCopy pairs are non-revertible and all sequence numbers are equal. In this
case, you do not need to take any action. Nevertheless, and depending on the failure
point-in-time, you may have to perform one of the above recovery actions. After this action, all
FlashCopy pairs will be non-revertible and all sequence numbers will be equal. Now you can
proceed to the next step.
280 IBM System Storage DS8000 Series: Copy Services in Open Environments
The next step then is to create the same consistency on the B volumes as we have on the C
volumes; see Figure 21-12. This can be achieved with the reverseflash command, with the
parameter -fast. This operation is called Fast Reverse Restore (FRR). You have to add the
-tgtpprc parameter with the reverseflash -fast command because the B volume is also the
Global Copy source at this step.
Server for
GM recovery
FRR FlashCopy
A
A A
Primary
Primary
A
Primary B
Primary
Primary
Primary
Primary
A
A
Primary GC source
FC source A
Primary
Primary
source C
Primary
Primary
suspended tertiary
pending
FC target
Though the Fast Reverse Restore operation starts the background copy from C to B volumes,
in the reverseflash command you have to specify the B volumes as the FlashCopy sources
and the C volumes as the FlashCopy targets. With the reverseflash command you have to
use the following parameters:
-fast
With this parameter, the reverseflash command can be issued before the background
copy completes. This option is intended for use as part of Global Mirror.
-tgtpprc
After the Failover B to A operation described in 21.5.3, “Global Copy Failover from B to A”
on page 276, the B volume became a Global Copy source volume in suspended state.
The -tgtpprc parameter allows the FlashCopy target volume to be a Global Copy source
volume. You have to specify this parameter here because the B volume becomes a
FlashCopy target in the reverseflash process.
Because you do not specify the -persist parameter, the FlashCopy relationship ends after the
background copy from C to B completes.
The above Fast Reverse Restore (FRR) operation does a background copy of all tracks that
changed on B since the last Consistency Group (CG) formation. This results in the B volume
becoming equal to the image that was present on the C volume. This is the logical view. From
the physical data placement point of view, the C volume does not have meaningful data after
the FlashCopy relationship ends.
You have to wait until all Fast Reverse Restore operations complete successfully and its
background copy completes before you proceed with the next step. Again, when the
background copy completes, the FlashCopy relation will end. Therefore, you should check if
the FlashCopy relationships remain to determine when all Fast Reverse Restore operations
are completed.
In a disaster situation it may be that you do not want to use the -nocp option for the FlashCopy
from B to C. This will remove the FlashCopy I/O overhead when the application starts.
Now you may restart the applications at the remote site using the B volumes. Note the B
volumes are Global Copy source volumes in suspended state, which implies that change
recording takes place. Later this allows you to re-synchronize from B to A, before returning to
the local site.
Eventually, here you may also decide to create another copy of this consistent set of volumes,
and create D volumes, or preserve this set of consistent volumes on tape.
When you create the D volumes, this is just a plain FlashCopy command. Note that this is a
full volume FlashCopy. This is because at a previous step we re-established the FlashCopy
relationship between B and C, indicating it was part of a Global Mirror environment. This
indication carries the start change recording attribute, which implies that you can only use the
same source volume for full volume FlashCopy. Only a single incremental relationship is
allowed per source volume.
Application
Server
Restart applications
a
Global Copy A
A A
Primary
Primary
Primary
A
Primary B
Primary
A
Primary
Primary
A
Primary source A
Primary
Primary
Primary
source suspended C
Primary
tertiary
pending
Note the suspended state of the B volumes, which implies change recording and indicates the
changed tracks on the B volumes. When the local site is about to become ready to restart the
applications again and resume the operations, you prepare the remote site for the next step.
282 IBM System Storage DS8000 Series: Copy Services in Open Environments
21.5.8 Prepare to switch back to the local site
The local site is back and operative again. We assume that the local site did not lose the data
at the time when the swap to the remote site was done. It is then possible to re-synchronize
the changed data from B to A before re-starting the application at the local site. This is
accomplished by doing a Failback operation (Copy Services Failback function) from B to A;
see Figure 21-14.
Application
Server
Failback B to A
re-synchronize from B to A
A
A A
Primary
Primary
A
Primary
Primary
Primary
Primary
B
Primary
A
A
Primary
source A
Primary
target Global Copy
copy pending
Primary
Primary
C
Primary a
copy pending
tertiary
Note that the Failback operation is issued to the B volumes as the source and the A volumes
as the target. This command changes the A volume from its previous source copy pending
state to target copy pending and starts the re-synchronization of the changes from B to A.
Note: Before doing the Failback operation, ensure that paths are defined from the remote
site LSS to its corresponding LSS at the local site.
Note that with Fibre Channel links you can define paths in either direction on the very same
FCP link. During the Failback operation the application stays running at the remote site to
minimize the application outage.
If the A volume is still online to the server at the local site or it was online when a crash
happened, so that the SCSI persistent reserve is still set on the previous source disk (A
volume), the Global Copy Failback process with the failbackpprc command fails. In this case
the server at the production site locks the target with a SCSI persistent reserve. After this
SCSI persistent reserve is reset with the varyoffvg command (in this case on AIX), the
failbackpprc command completes successfully.
There is a -resetreserve parameter for the failbackpprc command. This option resets the
reserved state so that the failback operation can complete. In the Failback operation after a
real disaster, you may use this parameter because the server might go down while the SCSI
persistent reserve was set on the A volume. In the planned failback operation, you must not
use this parameter because the server at the local site still owns the A volume and may be
using it and the Failback operation suddenly changes the contents of the volume. This may
cause the server file system’s corruption.
Failover A to B
A
A A
Primary
Primary
A
Primary
Primary
Primary
Primary
B
Primary
A
A
Primary
source Global Copy
source A
Primary
Primary
Primary
copy pending C
Primary
suspended
tertiary
Figure 21-15 shows the action at the local site, the Global Copy Failover operation from A to
B. This failoverpprc command will change the state of the A volumes from target copy
pending to source suspended, and start to keep a bitmap record of the changes to the A
volumes. You issue this command to the storage disk subsystem at the local site. The state of
the B volumes does not change.
Once the failover is completed, a failback operation from A to B is run; see Figure 21-16.
Failback A to B
Re-synchronize from A to B
A
A A
Primary
Primary
A
Primary
Primary
Primary
Primary
B
Primary
A
A
Primary
target
A
Primary
source Global Copy Primary
Primary
copy pending
copy pending C
Primary
tertiary
284 IBM System Storage DS8000 Series: Copy Services in Open Environments
Note: Before doing the Failback operation, ensure that paths are defined from the local site
LSS to its corresponding LSS at the remote site.
Figure 21-16 on page 284 shows the Failback operation at the local site. The failbackpprc
command will change the state of the A volumes from source suspended to source copy
pending. The state of the B volumes will change from source copy pending to target copy
pending. Also, the replication of updates from A to B begins. This replication ends quickly
because the application did not start yet at the local site.
Finally, if you did not already establish the FlashCopy relationships from B to C during the
Failover-Failback sequence at the remote site, then you have to do it now. This may be an
inband FlashCopy, as shown in Figure 21-17.
Server for
GM recovery
Remote FlashCopy B to C
FlashCopy
A
A A
Primary
Primary
A
Primary
Primary
Primary
Primary
B
Primary
A
A
Primary
source Global Copy
target
A
Primary
Primary
Primary
copy pending
copy pending C
Primary
tertiary
FlashCopy target
Local site Remote site
Figure 21-17 Establish Global Mirror FlashCopy relationship between B and C
The last step is to start the Global Mirror session again, as shown in Figure 21-18. Then the
application may resume at the local site.
FlashCopy
A
A A
Primary
Primary
A
Primary
Primary
Primary
Primary
B
Primary
A
A
Primary
source Global Copy
target
A
Primary
Primary
Primary
copy pending
copy pending C
Primary
Tertiary
FlashCopy target
Local site Remote site
Figure 21-18 Start Global Mirror session and resume application I/O at local site
In particular, the check on a valid Consistency Group after a production site failure is a
challenge when you consider a large configuration with many volume pairs. Each command
usually addresses a single pair of volumes. When you have a large quantity of volume pairs,
this requires automation, for example, to check all the FlashCopy relationships after a
production site failure. Note that for a planned site swap, you may not have to check for a valid
Consistency Group because, through a proper command sequence and verifying their
successful completion, the possibility of an inconsistent group of C volumes is very minimal, if
at all.
Global Mirror is a 2-site solution that can bridge any distance between both sites. There are
ready-to-use packages and services available to implement a disaster recovery solution for
2-site remote copy configurations. There is eRCMF that supports not only Global Copy and
Metro Mirror based configurations, but also Global Mirror configurations.
286 IBM System Storage DS8000 Series: Copy Services in Open Environments
22
The information discussed in this chapter can be complemented with the following sources:
Part 2, “Interfaces” on page 15
Chapter 8, “FlashCopy interfaces” on page 63
Chapter 17, “Global Copy interfaces” on page 201
The examples presented in Chapter 24, “Global Mirror examples” on page 305
The publication IBM System Storage DS8000 Command-Line Interface User´s Guide,
SC26-7916
This chapter gives an overview of the DS CLI and the DS SM for Global Mirror management.
DS CLI and DS SM are also covered in Part 2, “Interfaces” on page 15.
eRCMF is a scalable and flexible solution to protect business data and maintain consistent
data, providing an automation tool that simplifies the disaster recovery operations. For more
information about this solution see Part 9, “Solutions” on page 537.
288 IBM System Storage DS8000 Series: Copy Services in Open Environments
Task DS CLI DS SM
The DS CLI commands are issued via the Ethernet network to the DS Storage Management
Console (DS HMC). The DS CLI server resides on the DS HMC, which works as a gateway
between the clients and the storage disk subsystem. You can install a DS CLI client to access
the DS CLI server on various operating systems. See 14.2, “Copy Services network
components” on page 135, for a detailed description of the DS HMC network environment.
The DS CLI commands that are used for managing a Global Mirror environment fall into the
following basic groups:
For paths definitions and removal
For Global Copy pairs creation, management, and removal
For FlashCopy pairs creation, management, and removal
For Global Mirror session definition, management, and removal
The first three groups in the previous list refer to the same commands you use to manage and
control Global Copy and FlashCopy environments. Still, when working in a Global Mirror
environment specific command parameters must be used for some of the tasks. How these
commands are used and the options that must be selected is covered in the discussions and
examples we present in this part of the book.
For the Global Mirror session itself and its definition, management, modification, and removal,
we have the following DS CLI commands:
showgmir displays detailed properties and performance figures for Global Mirror.
showgmiroos displays the number of out of synchronization (out of sync) tracks for the
session.
mkgmir starts Global Mirror.
rmgmir stops Global Mirror.
pausegmir pauses Global Mirror.
resumegmir resumes Global Mirror.
mksession opens a Global Mirror session on a LSS.
chsession allows you to modify a Global Mirror session on a LSS. You can add and
remove a volume for Global Mirror with this command.
rmsession removes an existing Global Mirror session on a LSS.
lssession displays a Global Mirror session on a LSS.
For most DS CLI commands, you will need to know some (or all) of the following information:
The serial number and device type of the source and target storage disk subsystem.
The World Wide Node name (WWNN) of the remote storage disk subsystem.
The LSS numbers of the source and target volumes.
The port IDs for source and target. Up to eight port pair IDs can be specified.
A full establishment of a Global Mirror environment using DS CLI commands may take a long
time, specially if you need to set up an environment involving many volumes on many LSSs
and in several storage disk subsystems. For the setup and management of large-scale
environments, eRCMF or TPC for Replication are more appropriate, especially if you want to
test disaster recovery scenarios or if you want to run a real disaster recovery plan.
Detailed examples of how to set up and manage a Global Mirror environment using DS CLI
commands can be found in Chapter 24, “Global Mirror examples” on page 305.
290 IBM System Storage DS8000 Series: Copy Services in Open Environments
The DS CLI commands are documented in the publication IBM System Storage DS8000
Command-Line Interface User´s Guide, SC26-7916.
The DS SM interface is user friendly. However, you cannot use it for automation activities, and
certain Copy Services functions are not supported from this interface.
People who have fewer skills or less confidence with the other interfaces can use the DS
Storage Manager interface. It is reasonably simple to use; however, it is usually slower than
the other interfaces, and you cannot save actions for later use.
For Global Mirror setup activities, you will start from the following DS Storage Manager
panels:
For paths definitions, start from the Paths panel under the Copy Services menu. You can
see this panel in Figure 24-16 on page 356.
To create Global Copy pairs, start from the Metro Mirror panel under the Copy Services
menu. You can see this panel in Figure 24-23 on page 359.
To create FlashCopy pairs, start from the FlashCopy panel under the Copy Services
menu. You can see this panel in Figure 24-30 on page 364.
For Global Mirror session setup and control, start from the Global Mirror panel under the
Copy Services menu of the DS SM; see Table 22-1 on page 288.
If you select any check box for a Global Mirror session, the pull-down list will allow session
creation and management. The pull-down list will look like Figure 22-3.
292 IBM System Storage DS8000 Series: Copy Services in Open Environments
Consider that a full establishment of a Global Mirror environment requires using each Copy
Services panel in the DS SM and, if you need to set up an environment involving many
volumes on many LSSs and in several storage disk subsystems, this can take a very long
time. Therefore, the DS SM is more convenient for specific non-large-scale Copy Services
activities, and also it is a very didactic tool for users not skilled in other interfaces.
For the setup and management of large-scale environments, eRCMF or TPC for Replication
are more appropriate, especially if you want to test disaster recovery scenarios or if you want
to run a real disaster recovery plan.
Detailed examples of how to set up and manage a Global Mirror environment using the DS
Storage Manager GUI can be found in Chapter 24, “Global Mirror examples” on page 305.
22.4 eRCMF
The enterprise Remote Copy Management Facility (eRCMF) can be used to manage Metro
Mirror, Global Copy, and Global Mirror functions, as well as their associated FlashCopy tasks.
eRCMF can be use to manage Copy Services functions in a mixed open systems and z/OS
environment. It is a powerful tool that can consolidate the equivalent of several DS CLI
commands with just one command. Nevertheless, it is not using DS CLI scripts.
eRCMF is convenient for the management of medium and large scale environments. You will
find eRCMF indispensable if you want to test disaster recovery scenarios or if you want to run
a real disaster recovery plan.
eRCMF allows the automation of the recovery activities. It will not only trap SNMP messages,
but it will also execute commands in order to shorten downtime.
For a detailed description of eRCMF, from its installation through its configuration to its usage,
see Chapter 36, “IBM TotalStorage Rapid Data Recovery for UNIX and Windows” on
page 575.
This section discusses the impact of Global Copy and FlashCopy in the overall performance
of Global Mirror. Global Copy has at most only a minimal impact to the response time of an
application write I/O to a Global Copy source volume.
In the Global Mirror environment FlashCopy is used with the nocopy attribute. This implies
that, for any write I/Os to the FlashCopy source volume, there will be additional internally
triggered I/Os in the remote storage disk subsystem. These I/Os preserve the FlashCopy
source volume tracks by making a copy to the FlashCopy target volume before the source
tracks are updated. This happens every time within the interval in between two FlashCopy
establish operations.
Note that FlashCopy with the nocopy option does not start a background copy operation, but
it only maintains a set of FlashCopy bitmaps for the B and C volumes. These bitmaps are
established the first time a FlashCopy relationship is created with the nocopy option. Before a
source track is modified between the creation of two Consistency Groups, the track is copied
to the target C volume to preserve the previous point-in-time copy. This includes updates to
the corresponding bitmaps to reflect the new location of the track, which belongs to the
point-in-time copy. Each Global Copy write to its target B volume within the window of two
adjacent Consistency Groups causes such a FlashCopy I/O operation. See Figure 23-1.
Figure 23-1 Global Copy with write hit at the remote site
An I/O in this context is also a Global Copy I/O when Global Copy replicates a track from a
Global Copy source volume to a Global Copy target volume. This implies an additional read
I/O in the source storage disk subsystem at the local site and a corresponding write I/O to the
target storage disk subsystem at the remote site. In a Global Mirror environment, the Global
296 IBM System Storage DS8000 Series: Copy Services in Open Environments
Copy target volume is also the FlashCopy source volume. This may have some effect on a
very busy disk subsystem.
Figure 23-1 on page 296 roughly summarizes what happens between two Consistency Group
creation points when application writes come in. The application write I/O completes
immediately at the local site (1). Eventually Global Copy replicates the application I/O and
reads the data at the local site (2). Because of the FlashCopy nocopy attribute, before the
track gets updated on the B volume it is first preserved on the C volume.
Note: What Figure 23-1 and Figure 23-2 show is not necessarily the exact sequence of
internal I/O events, but a logical view approximation. There are internal microcode
optimization and consolidation techniques that make the entire process more efficient.
What is shown in Figure 23-1 on page 296 is the normal sequence of I/Os within a Global
Mirror configuration. The write (3) to the B1 volume is usually a write hit in the persistent
memory (NVS) of the target disk subsystem, so this is an instant operation. Eventually at
some later moment the B1 track is copied from B1 to C1 so it can be preserved and the
update in NVS can be destaged to the B1 volume.
There is some potential impact on the Global Copy data replication operation depending on
whether persistent memory or non-volatile cache is overcommitted in the target storage disk
subsystem. This may cause the FlashCopy source tracks to have to first be preserved on the
target volume C1 before the Global Copy write is completed. See Figure 23-2.
Figure 23-2 Global Copy with overcommitted NVS at the remote site
Figure 23-2 roughly summarizes what happens when persistent memory or NVS in the
remote storage disk subsystem is overcommitted. A read (3) and a write (4) to preserve the
FlashCopy source track and write it to the C1 volume are required before the write (5) can
complete. Then, when the track is updated on the B1 volume, this completes the write (5)
operation. Nevertheless, what you should usually see is all quick writes to cache and
persistent memory as Figure 23-1 on page 296 outlines.
All write I/Os to FlashCopy source volumes also trigger bitmap updates. This is the bitmap
that was created when the FlashCopy volume pair was created with the start change
recording attribute. This allows the replication of only the changed recording bitmap to the
corresponding bitmap for the target volume in the course of forming a Consistency Group.
See 20.4, “Consistency Groups” on page 259.
Maximum Maximum
coordination drain
time time
Write
Hold write I/O
A2 B2
source Global Copy paths target C2
tertiary
The coordination time, which you can limit by specifying a number of milliseconds, is the
maximum impact to an application’s write I/Os that will be allowed when forming a
Consistency Group. The intention is to keep this time window as small as possible. The
default of 50 ms may be a bit high in a transaction processing environment. A valid number
may also be a very low number — in the single-digit range. The required communication
between the master storage disk subsystem and the subordinate disk subsystems is inband
over the paths between the master and the subordinates. This communication is highly
optimized and allows you to minimize the potential application write I/O impact to 3 ms, for
example. Note that this communication is performed over FCP links. There is at least one
FCP link required between a master storage disk subsystem and a subordinate. For
redundancy, we suggest using two FCP links.
The following example illustrates the impact of the coordination time when Consistency Group
formation starts, and whether this impact has the potential to be significant.
Assume a total aggregate number of 5000 write I/Os over two source disk subsystems with
2500 write I/Os per second to each disk subsystem. Each write I/O takes 0.5 ms. You
specified 3 ms maximum to coordinate between the master the subordinate disk subsystems.
Assume further that a Consistency Group is created every 3 seconds, which is a goal set with
the Consistency Group interval time of zero.
5000 write I/Os.
0.5 ms response time for each write I/O.
Maximum coordination time is 3 ms.
Every 3 seconds a Consistency Group is created.
This is 5 I/Os for every millisecond or 15 I/Os within 3 ms. So each of these 15 write I/Os
experience a 3 ms delay. This happens every 3 seconds. Then we observe an average
response time delay of approximately:
298 IBM System Storage DS8000 Series: Copy Services in Open Environments
(15 IOs * 0.003 sec) / 3*5000 IO/sec) = 0.000003 sec or 0.003 ms
This drain period may also be limited to a maximum drain time. The default is 30 seconds and
is considered to be too small in a potentially write-intensive workload. Consider a number in
the range of 300 seconds to 600 seconds.
This replication process usually does not impact the application write I/O. There is a very low
possibility for a track within a Consistency Group to be updated before this track is replicated
to the remote site within this drain time period. When this unlikely event happens, the track is
immediately replicated to the target disk subsystem before the application write I/O modifies
the original track. The involved application write I/O experiences a similar response time delay
as though the I/O had been written to a Metro Mirror source volume. Note that subsequent
writes to this same track do not experience any delay because the track has already been
replicated to the remote site.
Through a one-to-one mapping from a local to a remote disk subsystem you achieve the
same configuration at the remote site for the B volumes and the C volumes. Figure 23-4 on
page 300 proposes spreading the B and C volumes across different ranks.
Primary
A Primary
A Primary
B1
Primary C3
A1
Primary
Primary target tertiary
source
copy pending
copy pending
Rank 1 Rank 1
Primary
FCP links A Primary
A FCP FCP
Primary
B2
Primary C1
A2
Primary
Primary
target tertiary
source
copy pending
copy pending
Rank 2 Rank 2
Primary
A Primary
A Primary
B3
Primary C2
A3
Primary
Primary target tertiary
source
copy pending
Host copy pending
Rank 3 Rank 3
Local site Remote site
Figure 23-4 Remote disk subsystem with all ranks containing equal numbers of volumes
With mixed Disk Drive Module (DDM) capacities and different speeds at the remote storage
disk subsystem, you may consider spreading B volumes not only over the fast DDMs but over
all ranks. Basically, follow a similar approach, as Figure 23-4 recommends. You may keep
particularly busy B volumes and C volumes on the faster DDMs.
300 IBM System Storage DS8000 Series: Copy Services in Open Environments
Figure 23-5 shows a configuration that incorporates the D volumes, which you can create
once in a while for test or other purposes.
Primary
A Primary
A Primary
B1
Primary C3
A1
Primary
Primary target tertiary
source
copy pending
copy pending
Rank 1 Rank 1
Primary
FCP links A Primary
A FCP FCP
Primary
B2
Primary C1
A2
Primary
Primary
target tertiary
source
copy pending
copy pending
Rank 2 Rank 2
Primary
A Primary
A Primary
B3
Primary C2
A3
Primary
Primary target tertiary
source
copy pending
Host copy pending
Rank 3 Rank 3
Local site
Primary
A Primary
Primary
Host D1
Primary D2
Primary
A Primary
Remote Primary
D3
Primary D4
site
Rank 4
For a situation like the one illustrated in Figure 23-5 we suggest as an alternative, a rank with
larger and perhaps slower DDMs. The D volumes may be read from another host, and any
other I/O to the D volumes does not impact the Global Mirror volumes in the other ranks.
Note that if you use the nocopy option for the relationship between the B and D volumes, this
will read the data from B when read I/Os to the D volume happen. So for this circumstance
you may consider using the copy option, thus preventing additional I/O to the ranks with the B
volumes. However, in this case, until the background copy between B and D completes, there
may be some impact to the Global Mirror data transfer.
An option here may be to spread all B volumes across all ranks again and also configure the
same number of volumes in each rank. Still, put the B and C volumes in different ranks. We
further recommend that you configure corresponding B and C volumes in such a way that
these volumes have an affinity to the same server. It would be ideal to also have the B
volumes connected to a different DA pair than the C volumes.
Example 23-1 on page 302 shows showgmiroos command outputs, which display
OutOfSyncTracks for the storage image scope as well as two LSS scopes. The first
showgmiroos command with the -scope si parameter shows that some Out Of Sync Tracks
remain within the disk subsystem. The second showgmiroos command with the -scope lss
This is different with LSS 11, where there is still data that has not yet been transferred to the
remote site. The third showgmiroos command shows that there are 67,847 tracks or roughly
4.14 GBs still waiting to get replicated from the local to the remote site. This situation denotes
a significant write I/O skew between LSS 10 and LSS 11.
Example 23-2 shows the lspprc command that reports the Out Of Sync Tracks by volume
level. Here we see that two volumes in LSS 10 have no Out Of Sync Tracks and two volumes
in LSS 11 have some number of Out Of Sync Tracks.
This particular configuration has source and target FlashCopy volumes within the very same
rank. Either you may consider putting one of the two busy volumes in LSS 11 into LSS 10 so
as to distribute the load over both LSSs, or another approach is to configure the FlashCopy
source volume and its corresponding target into different ranks or LSSs. See 23.4, “Remote
storage disk subsystem configuration” on page 299.
When the load distribution is unknown in your configuration, you may consider developing
some rudimentary code based on a script, for example, which regularly issues the
showgmiroos commands (as shown in Example 23-1) and the lspprc -l command )as
shown in Example 23-2). You may then process the output of these commands to better
understand the write load distribution over the Global Copy source volumes.
Note that the numbers in Example 23-1 may be just a brief peak period. It is still feasible to
use the conventional approach with I/O performance reports, such as iostat in the UNIX
environment, to investigate the write workload. Or Tivoli Productivity Center for Disk could be
used to analyze the storage disk subsystem performance.
302 IBM System Storage DS8000 Series: Copy Services in Open Environments
23.6 Growth within Global Mirror configurations
When you add a rather large number of volumes at once to an existing Global Mirror session,
then the available resources for Global Copy within the affected ranks might be over-utilized
or even monopolized by the initial copy pass. To avoid too much impact, consider adding
many new volumes to an existing Global Mirror session in stages. If possible, and as a rough
rule of thumb, add only a few volumes in a rank during application peak I/O periods. Once the
first initial copy is complete, add the next few volumes. Again, as a rule of thumb, plan for
adding one or two volumes in a rank during peak I/O load for the first initial copy pass. If
possible, then plan a massive add of new Global Copy volumes into an existing session
during off-peak periods.
The information discussed in this chapter is complemented with the publication IBM System
Storage DS8000 Command-Line Interface User´s Guide, SC26-7916.
In this configuration, different LSS and LUN numbers were used across the A, B, and C
components, so that you could unmistakably identify every element when referenced along
the discussion.
Note: In a real environment, and differently from our example, to simplify the management
of your Global Mirror environment, it is better for you to maintain a symmetrical
configuration in terms of both physical and logical elements.
.
FlashCopy
Global Copy
A B C
LSS10 LSS20 FlashCopy LSS22
Global Copy paths
GM Master
1000 FCP link 2000 2200
1001 2001 2201
306 IBM System Storage DS8000 Series: Copy Services in Open Environments
24.1.3 Setup procedure
The sequence of steps for the creation of a Global Mirror environment is not completely fixed,
and allows for some variation. Still, we recommend that you follow this procedure:
1. Create Global Copy relationships (A to B volumes).
2. Create FlashCopy relationships (B to C volumes).
3. Start Global Mirror session.
This procedure is followed in Example 24-1 where you can see the sequence of commands
and the corresponding results.
Note that the tasks you must perform to create the Global Copy relationships in a Global
Mirror environment are similar to what has been presented in Part 5, “Global Copy” on
page 185. You can refer to 19.1.2, “Set up of Global Copy configuration” on page 215.
====================================================================================================================
<< wait to see that the Out Of Sync Tracks shows 0 >>
=====================================================================================================================
Depending on your network environment, you can give the FlashCopy command to the local
DS8000#1 for its inband transmission to the remote DS8000#2. In this case you use the
308 IBM System Storage DS8000 Series: Copy Services in Open Environments
mkremoteflash command. Alternatively, if you have connectivity to the remote DS8000#2,
then you can give the mkflash command directly to the DS8000#2.
In our example we use the inband functionality of FlashCopy, in which case we have to
specify the LSS having the A volume for the -conduit parameter and the storage image ID at
the remote site for the -dev parameter; see Example 24-2. You have to give this command to
the DS HMC connected to the local DS8000#1.
Because the -nocp parameter is specified and the Global Copy initial copy (first pass)
completed in the previous step, no FlashCopy background copy occurs at this time.
Note: You can create this FlashCopy relationship before the initial copy of Global Copy.
However, to avoid unnecessary FlashCopy background I/Os we do not recommend it.
In our example, we have two LSSs in the local DS8000 that are going to participate in the
Global Mirror environment, LSS10 and LSS11. Therefore, we give the mksession command
twice and in each occasion we use the -lss parameter to specify the selected LSS. You also
With the chsession command you specify the Global Mirror session ID (02 in our example)
and the volumes that are going to be part of the Global Mirror environment. After adding the A
volumes to the session, the lssession command shows the volumes ID with its state in the
310 IBM System Storage DS8000 Series: Copy Services in Open Environments
LSS. At this time of the procedure, the volumes state is join pending. This means that the
volumes are not active for the current session. We have not yet started the Global Mirror
session.
Note: At this step we do not have to do anything to add the B and C volumes to the Global
Mirror session. They are automatically recognized by the Global Mirror mechanism through
the Global Copy relationships and the FlashCopy relationships.
In addition to the chsession command, you can also add the A volumes with the mksession
command when you define the Global Mirror session on a LSS; see Example 24-5.
Example 24-5 Add the A volumes when you create a Global Mirror session
dscli> mksession -dev IBM.2107-7520781 -lss IBM.2107-7520781/10 -volume 1000-1001 02
Date/Time: November 10, 2005 2:16:50 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
CMUC00145I mksession: Session 02 opened successfully.
dscli>
dscli> mksession -dev IBM.2107-7520781 -lss IBM.2107-7520781/11 -volume 1100-1101 02
Date/Time: November 10, 2005 2:17:02 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
CMUC00145I mksession: Session 02 opened successfully.
dscli>
dscli> lssession -l IBM.2107-7520781/10
Date/Time: November 10, 2005 2:17:09 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
LSS ID Session Status Volume VolumeStatus PrimaryStatus SecondaryStatus FirstPassComplete AllowCascading
=================================================================================================================
10 02 Normal 1000 Join Pending Primary Copy Pending Secondary Simplex True Disable
10 02 Normal 1001 Join Pending Primary Copy Pending Secondary Simplex True Disable
dscli>
dscli> lssession -l IBM.2107-7520781/11
Date/Time: November 10, 2005 2:17:13 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
LSS ID Session Status Volume VolumeStatus PrimaryStatus SecondaryStatus FirstPassComplete AllowCascading
=================================================================================================================
11 02 Normal 1100 Join Pending Primary Copy Pending Secondary Simplex True Disable
11 02 Normal 1101 Join Pending Primary Copy Pending Secondary Simplex True Disable
In the mkgmir command, the LSS we specified with the -lss parameter becomes the master. In
our example, it is LSS10. With this command we also specify the Global Mirror session ID of
the session we are starting.
At the time when you start the Global Mirror session, you can change the Global Mirror tuning
parameters of the session with the mkgmir command:
-cginterval: Specifies how long to wait between the formation of Consistency Groups. If
this number is not specified or is set to zero, Consistency Groups are formed continuously.
-coordinate: Indicates the maximum time that Global Mirror processing can hold host I/Os
in the source disk subsystem to start forming a Consistency Group.
-drain: Specifies maximum amount of time in seconds allowed for the data to drain to the
remote site before failing the current Consistency Group.
For additional discussion about the tuning parameters refer to 20.4.2, “Consistency Group
parameters” on page 261 and Chapter 23, “Global Mirror performance and scalability” on
page 295.
This showgmir command shows the Global Mirror current status. The Copy State field
indicates Running, which means that Global Mirror is satisfactorily operating. A Fatal state
would have indicated that Global Mirror failed, and the Fatal Reason field would have shown
the reason for the failure.
The showgmir command also shows the current time in the Current Time field, which is the
time when the DS8000 received this command. The time when the last successful
Consistency Group was formed is shown in the CG Time field. You can obtain the current
Recovery Time Objective (RPO) for this Global Mirror session from the difference between
the Current Time and the CG Time.
From the lssession command in Example 24-7 you can see that after starting the Global
Mirror session, the VolumeStatus of the A volumes changes from Join Pending to Active.
Example 24-7 The A volumes status after starting the Global Mirror session
dscli> lssession 10-11
Date/Time: November 11, 2005 5:21:48 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
LSS ID Session Status Volume VolumeStatus PrimaryStatus SecondaryStatus FirstPassComplete AllowCascadin
========================================================================================================================
10 02 CG In Progress 1000 Active Primary Copy Pending Secondary Simplex True Disable
10 02 CG In Progress 1001 Active Primary Copy Pending Secondary Simplex True Disable
11 02 CG In Progress 1100 Active Primary Copy Pending Secondary Simplex True Disable
11 02 CG In Progress 1101 Active Primary Copy Pending Secondary Simplex True Disable
If we use the -metrics parameter with the showgmir command, we can obtain metrics for
Global Mirror after we have started the session; see Example 24-8.
312 IBM System Storage DS8000 Series: Copy Services in Open Environments
Last Successful CG Form Time 11/11/2005 17:28:49 JST
Coord. Time (seconds) 50
Interval Time (seconds) 0
Max Drain Time (seconds) 30
First Failure Control Unit -
First Failure LSS -
First Failure Status No Error
First Failure Reason -
First Failure Master State -
Last Failure Control Unit -
Last Failure LSS -
Last Failure Status No Error
Last Failure Reason -
Last Failure Master State -
Previous Failure Control Unit -
Previous Failure LSS -
Previous Failure Status No Error
Previous Failure Reason -
Previous Failure Master State -
In Example 24-8 on page 312, the Total Failed CG Count field indicates the total number of
Consistency Groups that did not complete successfully after we have started the Global
Mirror. The Total Successful CG Count indicates the total number of Consistency Groups that
completed successfully. First Failure indicates the first failure after we have started this
session. Last Failure indicates the latest failure, and Previous Failure indicates the failure
before the latest one. All this failure information will be cleared after we stop the Global Mirror
and start it again. Pausing and resuming the Global Mirror operation does not reset this
information.
Depending on the Global Mirror parameters you set and your system environment, the
Consistency Group formation can occasionally fail and the showgmir -metrics may show the
error messages. A typical case is that you see Max Drain Time Exceeded with the showgmir
command when data of the out-of-sync bitmap cannot be drained within the specified time.
However, this failure does not mean that you lose consistent data at the remote site, because
Global Mirror does not take FlashCopy (B to C) for the failed Consistency Group data. Global
Mirror will continue to attempt to form additional Consistency Groups without external
intervention. If failures repeatedly continue (no more Consistency Groups are formed), the
percentage of successful Consistency Groups is unacceptable (many failures occur), or the
frequency of Consistency Groups is not meeting your requirements (Recovery Point
Objective-RPO), then the failures are a problem and need to be addressed.
There is another command related to Global Mirror, which is the showgmiroos command. This
command reports the number of Out Of Sync Tracks that at a given moment Global Mirror has
to transmit to the remote site —the size of the logical track on the DS8000 FB volume is 64
KB. With the -scope parameter you select either the storage image scope or the LSS scope
for the information to be reported. See Example 24-9.
Figure 24-2 shows the example DS8000 configuration where we have a total of eight A
volumes (four on DS8000#1 and four DS8000#3) at the local site that participate in the Global
Mirror session. We have one DS8000 (DS8000#2) at the remote site with the corresponding
B and the C volumes.
314 IBM System Storage DS8000 Series: Copy Services in Open Environments
Example 24-10 shows how to start a Global Mirror configuration when there is a subordinate.
The example does not show how to set up the Global Copy and FlashCopy relationships
because these steps are exactly the same as in a no-subordinate situation.
<< Setup Global Mirror environment bewteen DS8000#3 and DS8000#2 (These steps are NOT shown here) >>
You can use the -quiet parameter to turn off the confirmation prompt for this command.
316 IBM System Storage DS8000 Series: Copy Services in Open Environments
Subordinate Count -
Master/Subordinate Assoc -
Example 24-11 on page 316 illustrates how to end Global Mirror session 02 processing.
Although this command might interrupt the formation of a consistency group, every attempt is
made to preserve the previous consistent copy of the data on the FlashCopy target volumes.
If, due to failures, this command cannot complete without compromising the consistent copy,
the command stops processing and an error code is issued. If this occurs, reissue the rmgmir
command with the -force parameter to force the command to stop the Global Mirror process.
If the Global Mirror configuration that you started also had subordinates, then to end Global
Mirror you also have to specify the Global Mirror control path information with the rmgmir
command. Otherwise, the command fails; see Example 24-12.
Example 24-13 Remove the A volumes from the Global Mirror session
dscli> lssession 10-11
Date/Time: November 10, 2005 10:21:16 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
LSS ID Session Status Volume VolumeStatus PrimaryStatus SecondaryStatus FirstPassComplete AllowCas
===========================================================================================================
10 02 Normal 1000 Join Pending Primary Copy Pending Secondary Simplex True Disable
10 02 Normal 1001 Join Pending Primary Copy Pending Secondary Simplex True Disable
11 02 Normal 1100 Join Pending Primary Copy Pending Secondary Simplex True Disable
11 02 Normal 1101 Join Pending Primary Copy Pending Secondary Simplex True Disable
dscli>
dscli> chsession -dev IBM.2107-7520781 -lss IBM.2107-7520781/10 -action remove -volume 1000-1001 02
Date/Time: November 10, 2005 10:21:29 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
CMUC00147I chsession: Session 02 successfully modified.
dscli>
dscli> chsession -dev IBM.2107-7520781 -lss IBM.2107-7520781/11 -action remove -volume 1100-1101 02
Date/Time: November 10, 2005 10:21:37 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
CMUC00147I chsession: Session 02 successfully modified.
dscli>
Example 24-14 Remove the Global Mirror session from the LSSs
dscli> lssession 10-11
Date/Time: November 10, 2005 10:25:45 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
LSS ID Session Status Volume VolumeStatus PrimaryStatus SecondaryStatus FirstPassComplete AllowCascading
========================================================================================================
10 02 - - - - - - -
11 02 - - - - - - -
dscli>
dscli> rmsession -quiet -dev IBM.2107-7520781 -lss IBM.2107-7520781/10 02
Date/Time: November 10, 2005 10:26:07 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
CMUC00146I rmsession: Session 02 closed successfully.
dscli>
dscli> rmsession -quiet -dev IBM.2107-7520781 -lss IBM.2107-7520781/11 02
Date/Time: November 10, 2005 10:26:11 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
CMUC00146I rmsession: Session 02 closed successfully.
dscli>
dscli> lssession 10-11
Date/Time: November 10, 2005 10:26:16 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
CMUC00234I lssession: No Session found.
Before closing the Global Mirror session on the LSS, we must remove all the A volumes from
the Global Mirror session on that LSS, otherwise, the rmsession command fails; see
Example 24-15.
Example 24-15 rmsession command fails when Global Mirror volumes were not previously removed
dscli> lssession 10-11
Date/Time: November 10, 2005 10:21:16 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
LSS ID Session Status Volume VolumeStatus PrimaryStatus SecondaryStatus FirstPassComplete AllowCas
===========================================================================================================
10 02 Normal 1000 Join Pending Primary Copy Pending Secondary Simplex True Disable
10 02 Normal 1001 Join Pending Primary Copy Pending Secondary Simplex True Disable
11 02 Normal 1100 Join Pending Primary Copy Pending Secondary Simplex True Disable
11 02 Normal 1101 Join Pending Primary Copy Pending Secondary Simplex True Disable
dscli>
dscli> rmsession -dev IBM.2107-7520781 -lss IBM.2107-7520781/10 02
Date/Time: November 10, 2005 10:25:14 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
CMUC00148W rmsession: Are you sure you want to close session 02? [y/n]:y
CMUN03107E rmsession: Copy Services operation failure: volumes in session
318 IBM System Storage DS8000 Series: Copy Services in Open Environments
24.2.4 Terminate FlashCopy pairs
Depending on your network environment, you can give the FlashCopy commands to the local
disk subsystem for its inband transmission to the remote disk subsystem. In this case you use
the rmremoteflash command. Alternatively, if you have connectivity to the remote disk
subsystem, then you can give the rmflash command directly to the remote disk subsystem.
In our example we use the inband functionality of FlashCopy, in which case we have to
specify the LSS having the A volume for the -conduit parameter and the storage image ID at
the remote site for the -dev parameter; see Example 24-16. You have to give this command to
the DS HMC connected to the local DS8000#1.
Example 24-16 Remove all FlashCopy relationships between the B and C volumes
dscli> lsremoteflash -conduit IBM.2107-7520781/10 -dev IBM.2107-75ABTV1 2000-2001
Date/Time: November 10, 2005 10:52:25 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-75ABTV1
ID SrcLSS SequenceNum ActiveCopy Recording Persistent Revertible SourceWriteEnabled TargetWriteEnabled Background
========================================================================================================================
2000:2200 20 43734829 Disabled Enabled Enabled Disabled Enabled Disabled Disabled
2001:2201 20 43734829 Disabled Enabled Enabled Disabled Enabled Disabled Disabled
dscli>
dscli> lsremoteflash -conduit IBM.2107-7520781/11 -dev IBM.2107-75ABTV1 2100-2101
Date/Time: November 10, 2005 10:52:33 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-75ABTV1
ID SrcLSS SequenceNum ActiveCopy Recording Persistent Revertible SourceWriteEnabled TargetWriteEnabled Background
========================================================================================================================
2100:2300 21 43734829 Disabled Enabled Enabled Disabled Enabled Disabled Disabled
2101:2301 21 43734829 Disabled Enabled Enabled Disabled Enabled Disabled Disabled
dscli>
dscli> rmremoteflash -quiet -conduit IBM.2107-7520781/10 -dev IBM.2107-75ABTV1 2000-2001:2200-2201
Date/Time: November 10, 2005 10:52:49 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-75ABTV1
CMUC00180I rmremoteflash: Removal of the remote FlashCopy volume pair 2000:2200 has been initiated successfully. Use the
lsremoteflash command to determine when the relationship is deleted.
CMUC00180I rmremoteflash: Removal of the remote FlashCopy volume pair 2001:2201 has been initiated successfully. Use the
lsremoteflash command to determine when the relationship is deleted.
dscli>
dscli> rmremoteflash -quiet -conduit IBM.2107-7520781/11 -dev IBM.2107-75ABTV1 2100-2101:2300-2301
Date/Time: November 10, 2005 10:53:49 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-75ABTV1
CMUC00180I rmremoteflash: Removal of the remote FlashCopy volume pair 2100:2300 has been initiated successfully. Use the
lsremoteflash command to determine when the relationship is deleted.
CMUC00180I rmremoteflash: Removal of the remote FlashCopy volume pair 2101:2301 has been initiated successfully. Use the
lsremoteflash command to determine when the relationship is deleted.
dscli>
dscli> lsremoteflash -conduit IBM.2107-7520781/10 -dev IBM.2107-75ABTV1 2000-2001
Date/Time: November 10, 2005 10:53:55 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-75ABTV1
CMUC00234I lsremoteflash: No Remote Flash Copy found.
dscli>
dscli> lsremoteflash -conduit IBM.2107-7520781/11 -dev IBM.2107-75ABTV1 2100-2101
Date/Time: November 10, 2005 10:53:59 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-75ABTV1
CMUC00234I lsremoteflash: No Remote Flash Copy found.
Note that the tasks you must perform to terminate the Global Copy relationships in a Global
Mirror environment are similar to what has been presented in Part 5, “Global Copy” on
Example 24-17 Remove all Global Copy pairs and remove the paths
dscli> lspprc 1000-1001 1100-1101
Date/Time: November 10, 2005 11:14:19 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
==================================================================================================
1000:2000 Copy Pending - Global Copy 10 unknown Disabled True
1001:2001 Copy Pending - Global Copy 10 unknown Disabled True
1100:2100 Copy Pending - Global Copy 11 unknown Disabled True
1101:2101 Copy Pending - Global Copy 11 unknown Disabled True
dscli>
dscli> rmpprc -remotedev IBM.2107-75ABTV1 -quiet 1000-1001:2000-2001 1100-1101:2100-2101
Date/Time: November 10, 2005 11:14:32 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
CMUC00155I rmpprc: Remote Mirror and Copy volume pair 1000:2000 relationship successfully withdrawn.
CMUC00155I rmpprc: Remote Mirror and Copy volume pair 1001:2001 relationship successfully withdrawn.
CMUC00155I rmpprc: Remote Mirror and Copy volume pair 1100:2100 relationship successfully withdrawn.
CMUC00155I rmpprc: Remote Mirror and Copy volume pair 1101:2101 relationship successfully withdrawn.
dscli>
dscli> lspprcpath 10-11
Date/Time: November 10, 2005 11:16:01 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
Src Tgt State SS Port Attached Port Tgt WWNN
=========================================================
10 20 Success FF20 I0143 I0010 5005076303FFC663
10 20 Success FF20 I0213 I0140 5005076303FFC663
11 21 Success FF21 I0143 I0010 5005076303FFC663
11 21 Success FF21 I0213 I0140 5005076303FFC663
dscli>
dscli> rmpprcpath -quiet -remotedev IBM.2107-75ABTV1 10:20 11:21
Date/Time: November 10, 2005 11:16:34 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
CMUC00150I rmpprcpath: Remote Mirror and Copy path 10:20 successfully removed.
CMUC00150I rmpprcpath: Remote Mirror and Copy path 11:21 successfully removed.
Note: The pausegmir command does not influence the Global Copy data transfer.
320 IBM System Storage DS8000 Series: Copy Services in Open Environments
When you pause a Global Mirror session with the pausegmir command it will complete the
Consistency Group formation in progress before it pauses. This is slightly different from the
usage of the rmgmir command; refer to 24.3.3, “Stop and start Global Mirror” on page 324.
The Status shown by the lssession command changes from CG In Progress, which means
that the Consistency Group of the session is in progress, to Normal, which means that the
session is in a normal Global Copy state. In fact, you see this state (Normal) between the
time when a FlashCopy has been taken and the next Global Copy Consistency Group
formation time.
Note: The FlashCopy sequence number has not changed after pausing Global Mirror
because the FlashCopy at the remote site has never been done; see Example 24-18.
The resumegmir command resumes Global Mirror processing for a specified session; see
Example 24-19. Consistency Group formation is resumed.
322 IBM System Storage DS8000 Series: Copy Services in Open Environments
Current Time 11/11/2005 01:20:41 JST
CG Time 11/11/2005 01:20:41 JST
Successful CG Percentage 100
FlashCopy Sequence Number 0x43737359
Master ID IBM.2107-7520781
Subordinate Count 0
Master/Subordinate Assoc -
Example 24-20 Change the Consistent Group interval (CG Interval) time parameter
dscli> showgmir -dev IBM.2107-7520781 IBM.2107-7520781/10
Date/Time: November 11, 2005 1:15:50 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID IBM.2107-7520781/10
Master Count 1
Master Session ID 0x02
Copy State Running
Fatal Reason Not Fatal
CG Interval (seconds) 0
XDC Interval(milliseconds) 50
CG Drain Time (seconds) 30
Current Time 11/11/2005 01:20:41 JST
CG Time 11/11/2005 01:20:41 JST
Successful CG Percentage 100
FlashCopy Sequence Number 0x43737359
Master ID IBM.2107-7520781
Subordinate Count 0
Master/Subordinate Assoc -
dscli>
dscli> pausegmir -dev IBM.2107-7520781 -lss IBM.2107-7520781/10 -session 02
Date/Time: November 11, 2005 1:23:10 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
CMUC00163I pausegmir: Global Mirror for session 02 successfully paused.
dscli>
dscli> resumegmir -dev IBM.2107-7520781 -lss IBM.2107-7520781/10 -session 02 -cginterval 60
Date/Time: November 11, 2005 1:23:35 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
CMUC00164I resumegmir: Global Mirror for session 02 successfully resumed.
dscli>
dscli> showgmir -dev IBM.2107-7520781 IBM.2107-7520781/10
Date/Time: November 11, 2005 1:23:39 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID IBM.2107-7520781/10
Master Count 1
Master Session ID 0x02
Copy State Running
Fatal Reason Not Fatal
CG Interval (seconds) 60
XDC Interval(milliseconds) 50
CG Drain Time (seconds) 30
Current Time 11/11/2005 01:28:29 JST
CG Time 11/11/2005 01:28:25 JST
Successful CG Percentage 100
FlashCopy Sequence Number 0x43737529
Master ID IBM.2107-7520781
Subordinate Count 0
Master/Subordinate Assoc -
You do not need to remove the Global Copy and FlashCopy relationships to stop the Global
Mirror master. After stopping the Global Mirror master, the Consistency Group formation does
not continue, which means the FlashCopy sequence number at the remote site does not
increment.
Although the operation to stop Global Mirror with the rmgmir command might interrupt the
formation of a Consistency Group, every attempt is made to preserve the previous consistent
copy of the data on the FlashCopy target volumes. If due to failures the rmgmir command
cannot complete without compromising the consistent copy, the command stops processing
and an error code is issued. If this occurs, reissue the rmgmir command with the -force
parameter to force the command to stop the Global Mirror process.
The mkgmir command restarts the Global Mirror master. You can specify another LSS on
which you will start the Global Mirror master. In Example 24-21, we stop the Global Mirror
master running on LSS10 and then start again on LSS11.
324 IBM System Storage DS8000 Series: Copy Services in Open Environments
11 02 Normal 1100 Active Primary Copy Pending Secondary Simplex True Disable
11 02 Normal 1101 Active Primary Copy Pending Secondary Simplex True Disable
dscli>
dscli> mkgmir -dev IBM.2107-7520781 -lss IBM.2107-7520781/11 -session 02
Date/Time: November 11, 2005 9:43:30 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
CMUC00162I mkgmir: Global Mirror for session 02 successfully started.
dscli>
dscli> lssession 10-12
Date/Time: November 11, 2005 9:43:35 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
LSS ID Session Status Volume VolumeStatus PrimaryStatus SecondaryStatus FirstPassComplete AllowCascading
========================================================================================================================
10 02 CG In Progress 1000 Active Primary Copy Pending Secondary Simplex True Disable
10 02 CG In Progress 1001 Active Primary Copy Pending Secondary Simplex True Disable
11 02 CG In Progress 1100 Active Primary Copy Pending Secondary Simplex True Disable
11 02 CG In Progress 1101 Active Primary Copy Pending Secondary Simplex True Disable
dscli>
dscli> showgmir -dev IBM.2107-7520781 IBM.2107-7520781/11
Date/Time: November 11, 2005 9:43:46 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID IBM.2107-7520781/11
Master Count 1
Master Session ID 0x02
Copy State Running
Fatal Reason Not Fatal
CG Interval (seconds) 0
XDC Interval(milliseconds) 50
CG Drain Time (seconds) 30
Current Time 11/11/2005 21:49:09 JST
CG Time 11/11/2005 21:49:08 JST
Successful CG Percentage 100
FlashCopy Sequence Number 0x43749344
Master ID IBM.2107-7520781
Subordinate Count 0
Master/Subordinate Assoc -
To be added to a Global Mirror session the A volumes may be in any state, such as simplex
(no relationship), copy pending, or suspended. Volumes that have not completed their initial
copy phase (also called first pass) stay in a Join Pending state until the first pass is complete.
You can check the first pass status with the lspprc -l command; see Example 24-23. The
First Pass Status field reports this information, where True means that the Global Copy first
pass has been completed.
Example 24-23 Check the first pass completion for the Global Copy initial copy
dscli> lspprc -l -fmt stanza 1002
Date/Time: November 11, 2005 9:04:50 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID 1002:2002
State Copy Pending
Reason -
Type Global Copy
Out Of Sync Tracks 0
Tgt Read Disabled
Src Cascade Disabled
Tgt Cascade invalid
Date Suspended -
SourceLSS 10
326 IBM System Storage DS8000 Series: Copy Services in Open Environments
Timeout (secs) unknown
Critical Mode Disabled
First Pass Status True
When you want to remove an A volume from the Global Mirror environment, you can use the
chsession -action remove -volume command. You first remove the A volume from the
Global Mirror session and then remove its Global Copy and FlashCopy relationships. See
Example 24-24.
Attention: Suspending or removing even one Global Copy pair that belongs to an active
Global Mirror session will impact the formation of Consistency Groups. If you suspend or
remove the Global Copy relationship from the A volume without removing the volume from
the Global Mirror session, it will cause Consistency Group formation to fail, and periodical
SNMP alerts will be issued.
After this, you add the LSS with the mksession command, and then you add the A volume.
You can also use the mksession command to add the LSS along with the A volume; see
Example 24-25.
328 IBM System Storage DS8000 Series: Copy Services in Open Environments
When you remove an LSS from a Global Mirror environment, you have to first remove all the
A volumes on the LSS with the chsession command and then remove the LSS with the
rmsession command; see Example 24-26.
This task is in fact a topology change of the Global Mirror configuration, which requires that
you stop Global Mirror first in order to re-start it again with the new configuration; see 21.3.4,
“Global Mirror environment topology changes” on page 269.
For examples of Global Mirror stop and start tasks see 24.3.3, “Stop and start Global Mirror”
on page 324.
For this example we use the configuration that we set up in 24.1, “Set up a Global Mirror
environment using the DS CLI” on page 306.
A B C
GC: Source GC: Target FC: Target
Copy Pending Copy Pending
FC: Source
DS8000#1 DS8000#2
-dev IBM.2107-7520781 -dev IBM.2107-75ABTV1
Figure 24-3 Global Mirror example before unplanned production site failure
The following sections discuss each of the listed tasks of the recovery scenario.
330 IBM System Storage DS8000 Series: Copy Services in Open Environments
Example 24-27 Terminate Global Mirror
dscli> rmgmir -dev IBM.2107-7520781 -lss IBM.2107-7520781/10 -session 02
Date/Time: November 11, 2005 11:28:47 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
CMUC00166W rmgmir: Are you sure you want to stop the Global Mirror session 02:? [y/n]:y
CMUC00165I rmgmir: Global Mirror for session 02 successfully stopped.
This Failover operation sets the stage for change recording when application updates start
changing the B volumes. Change recording in turn allows you to re-synchronize just the
changes from the B to the A volumes, later before returning to the local site. But at this stage,
the B volumes do not contain consistent data and are still useless. We just changed their
Global Copy state from target to source and suspended. Figure 24-4 shows the DS8000
environment after the Failover operation.
A B C
GC: Source GC: Source FC: Target
Copy Pending Suspended
FC: Source
DS8000#1 DS8000#2
-dev IBM.2107-7520781 -dev IBM.2107-75ABTV1
Example 24-28 shows the command for this operation. You can check the result with the
lspprc command.
Each FlashCopy pair needs a FlashCopy query to identify its state. You use the lsflash
command to check the SequenceNum and the Revertible field status. Example 24-29 shows
the Revertible status of all of the FlashCopy pairs is Disabled (that is, non-revertible) and the
SequenceNums of all relationships are equal. Therefore, you do not need to take any action in
this case.
You can find a detailed discussion of this verification process in 21.5.4, “Verify for valid
Consistency Group state” on page 277.
Example 24-30 shows a hypothetical situation where all FlashCopy relationships are in the
revertible state and have the same sequence number. In this case, you should execute a
revertflash to all the FlashCopy relationships.
332 IBM System Storage DS8000 Series: Copy Services in Open Environments
2101:2301 21 437895B2 300 Disabled Enabled Enabled Enabled Disabled Disabled Disabled
dscli>
dscli> revertflash 2000-2001 2100-2101
Date/Time: November 14, 2005 11:21:09 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-75ABTV1
CMUC00171I revertflash: FlashCopy volume pair 2000:2000 successfully reverted.
CMUC00171I revertflash: FlashCopy volume pair 2001:2001 successfully reverted.
CMUC00171I revertflash: FlashCopy volume pair 2100:2100 successfully reverted.
CMUC00171I revertflash: FlashCopy volume pair 2101:2101 successfully reverted.
dscli>
dscli> lsflash 2000-2001 2100-2101
Date/Time: November 14, 2005 11:21:11 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-75ABTV1
ID SrcLSS SequenceNum Timeout ActiveCopy Recording Persistent Revertible SourceWriteEnabled TargetWriteEnabled BackgroundCopy
====================================================================================================================================
2000:2200 20 437895B2 300 Disabled Enabled Enabled Disabled Enabled Disabled Disabled
2001:2201 20 437895B2 300 Disabled Enabled Enabled Disabled Enabled Disabled Disabled
2100:2300 21 437895B2 300 Disabled Enabled Enabled Disabled Enabled Disabled Disabled
2101:2301 21 437895B2 300 Disabled Enabled Enabled Disabled Enabled Disabled Disabled
dscli>
If some FlashCopy pairs are revertible and others are not revertible while their sequence
numbers are equal, you should execute a commitflash command to the FlashCopy
relationships that have the revertible status; see Example 24-31.
When the FlashCopy relationship is not in a revertible state, the commit operation is not
possible. When you give this command to FlashCopy pairs that are non-revertible, you are
going to see only an error message, but no action is performed. To make the task easier, you
may run a commitflash command to all FlashCopy pairs. In Example 24-31, 2000 and 2001
are not in the revertible state; therefore, we see error messages.
Example 24-31 Some are revertible and SeqNum are equal, then commitflash
dscli> lsflash 2000-2001 2100-2101
Date/Time: November 14, 2005 11:18:21 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-75ABTV1
ID SrcLSS SequenceNum Timeout ActiveCopy Recording Persistent Revertible SourceWriteEnabled TargetWriteEnabled BackgroundCopy
====================================================================================================================================
2000:2200 20 437895B2 300 Disabled Enabled Enabled Disabled Enabled Disabled Disabled
2001:2201 20 437895B2 300 Disabled Enabled Enabled Disabled Enabled Disabled Disabled
2100:2300 21 437895B2 300 Disabled Enabled Enabled Enabled Disabled Disabled Disabled
2101:2301 21 437895B2 300 Disabled Enabled Enabled Enabled Disabled Disabled Disabled
dscli>
dscli> commitflash 2000-2001 2100-2101
Date/Time: November 14, 2005 11:18:56 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-75ABTV1
CMUN03054E commitflash: 2000:2000: Copy Services operation failure: invalid revertible specification
CMUN03054E commitflash: 2001:2001: Copy Services operation failure: invalid revertible specification
CMUC00170I commitflash: FlashCopy volume pair 2100:2100 successfully committed.
CMUC00170I commitflash: FlashCopy volume pair 2101:2101 successfully committed.
dscli>
dscli> lsflash 2000-2001 2100-2101
Date/Time: November 14, 2005 11:19:04 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-75ABTV1
ID SrcLSS SequenceNum Timeout ActiveCopy Recording Persistent Revertible SourceWriteEnabled TargetWriteEnabled BackgroundCopy
====================================================================================================================================
2000:2200 20 437895B2 300 Disabled Enabled Enabled Disabled Enabled Disabled Disabled
2001:2201 20 437895B2 300 Disabled Enabled Enabled Disabled Enabled Disabled Disabled
2100:2300 21 437895B2 300 Disabled Enabled Enabled Disabled Enabled Disabled Disabled
2101:2301 21 437895B2 300 Disabled Enabled Enabled Disabled Enabled Disabled Disabled
Example 24-32 on page 334 shows a hypothetical situation where some FlashCopy pairs are
revertible and others are non-revertible. The revertible FlashCopy pairs' sequence numbers
are equal. The non-revertible FlashCopy pairs’ sequence numbers are also equal, but do not
match the revertible FlashCopies sequence number. In this case, you should give a
commitflash command to the FlashCopy relationships that have the revertible status.
When the FlashCopy relationship is non-revertible, the revert operation is not possible. When
you execute this command against FlashCopy pairs that are non-revertible, you are going to
see only an error message, but no action is performed. To make the task easier, you may run
Example 24-32 Some are revertible and SeqNum are not equal, then revertflash
dscli> lsflash 2000-2001 2100-2101
Date/Time: November 14, 2005 10:49:47 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-75ABTV1
ID SrcLSS SequenceNum Timeout ActiveCopy Recording Persistent Revertible SourceWriteEnabled TargetWriteEnabled BackgroundCopy
====================================================================================================================================
2000:2200 20 437895B2 300 Disabled Enabled Enabled Disabled Enabled Disabled Disabled
2001:2201 20 437895B2 300 Disabled Enabled Enabled Disabled Enabled Disabled Disabled
2100:2300 21 437895B3 300 Disabled Enabled Enabled Enabled Disabled Disabled Disabled
2101:2301 21 437895B3 300 Disabled Enabled Enabled Enabled Disabled Disabled Disabled
dscli>
dscli> revertflash 2000-2001:2200-2201 2100-2101:2300-2301
Date/Time: November 14, 2005 11:13:28 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-75ABTV1
CMUC00083E revertflash: Invalid volume 2000-2001:2200-2201.
dscli>
dscli> revertflash 2000-2001 2100-2101
Date/Time: November 14, 2005 11:14:14 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-75ABTV1
CMUN03054E revertflash: 2000:2000: Copy Services operation failure: invalid revertible specification
CMUN03054E revertflash: 2001:2001: Copy Services operation failure: invalid revertible specification
CMUC00171I revertflash: FlashCopy volume pair 2100:2100 successfully reverted.
CMUC00171I revertflash: FlashCopy volume pair 2101:2101 successfully reverted.
dscli>
dscli> lsflash 2000-2001 2100-2101
Date/Time: November 14, 2005 11:14:25 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-75ABTV1
ID SrcLSS SequenceNum Timeout ActiveCopy Recording Persistent Revertible SourceWriteEnabled TargetWriteEnabled BackgroundCopy
====================================================================================================================================
2000:2200 20 437895B2 300 Disabled Enabled Enabled Disabled Enabled Disabled Disabled
2001:2201 20 437895B2 300 Disabled Enabled Enabled Disabled Enabled Disabled Disabled
2100:2300 21 437895B2 300 Disabled Enabled Enabled Disabled Enabled Disabled Disabled
2101:2301 21 437895B2 300 Disabled Enabled Enabled Disabled Enabled Disabled Disabled
After these actions, all FlashCopy pairs are non-revertible and all sequence numbers are
equal, so now you can proceed to the next step.
We want to have to two good copies of the data at the recovery site. The aim is to have a
consistent set of volumes to work with, still keeping a good copy to which we can resort to if
needed. The next step is then to create the same consistency on the B volumes as we have
on the C volumes; see Figure 24-5 on page 335. This can be achieved with the reverseflash
-fast command. This operation is called Fast Reverse Restore (FRR). You have to use the
-tgtpprc parameter with the reverseflash -fast command because the B volume is also a
Global Copy source at this step.
Note: Though the Fast Reverse Restore operation starts a background copy from the C to
the B volumes, in the reverseflash command you have to specify the B volumes as the
FlashCopy sources and the C volumes as the FlashCopy targets.
334 IBM System Storage DS8000 Series: Copy Services in Open Environments
Figure 24-5 shows the remote DS8000 environment after the reverseflash command was
issued. After executing this command and before C to B background copy is completed, the C
volumes become the FlashCopy source and the B volumes become the FlashCopy target.
A B C
GC: Source GC: Source FC: Source
Copy Pending Suspended
FC: Target
DS8000#1 DS8000#2
-dev IBM.2107-7520781 -dev IBM.2107-75ABTV1
Figure 24-5 Site swap scenario after the reverseflash command was issued
Example 24-33 shows the results of the reverseflash command. The lsflash command
shows volume 2200 (C volume) as the FlashCopy source.
Because you do not specify the -persist parameter, the FlashCopy relationship ends after the
background copy from C to B completes; see Figure 24-6.
A B C
GC: Source GC: Source FC: None
Copy Pending Suspended
FC: None
DS8000#1 DS8000#2
-dev IBM.2107-7520781 -dev IBM.2107-75ABTV1
You have to wait until all Fast Reverse Restore operations and their background copy
complete successfully before you proceed with the next step. Again, when the background
copy completes, the FlashCopy relation will end. Therefore, you should check if the
FlashCopy relationships remain to determine when all Fast Reverse Restore operations have
completed; see Example 24-34. This example shows the result of the lsflash command after
the reverseflash background copy completes.
Example 24-34 The lsflash command to confirm the completion of the background copy
dscli> lsflash 2000-2001 2100-2101
Date/Time: November 11, 2005 11:57:17 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-75ABTV1
CMUC00234I lsflash: No Flash Copy found.
336 IBM System Storage DS8000 Series: Copy Services in Open Environments
24.4.5 Re-establish the FlashCopy relationship from B to C
LSS10 LSS20
FlashCopy LSS22
A B C
GC: Source GC: Source FC: Target
Copy Pending Suspended
FC: Source
DS8000#1 DS8000#2
-dev IBM.2107-7520781 -dev IBM.2107-75ABTV1
In this step you create the former FlashCopy relationship between the B and C volumes, as
they were at the beginning when you set up the Global Mirror environment; see Figure 24-7.
This step is in preparation for returning later to production at the local site. The mkflash
command used in this step is illustrated in Example 24-35, and is exactly the same
FlashCopy command you may have used when you initially created the Global Mirror
environment. See 20.3.4, “Introduce FlashCopy” on page 256.
In a disaster situation it may be that you do not want to use the -nocp option for the FlashCopy
from B to C. This will remove the FlashCopy I/O overhead when the application starts.
Now that the application has started at the remote site, all the write I/Os to the new source
volumes (that is, the B volumes) are tracked in the bit-maps by the Failover function.
Figure 24-8 shows this environment.
I/O
LSS10 LSS20
FlashCopy LSS22
A B C
GC: Source GC: Source FC: Target
Copy Pending Suspended
FC: Source
DS8000#1 DS8000#2
-dev IBM.2107-7520781 -dev IBM.2107-75ABTV1
338 IBM System Storage DS8000 Series: Copy Services in Open Environments
24.5.1 Create paths from B to A
The local site is operational again. If the local site did not lose the data at the time when the
swap to the remote site occurred, then it is possible to re-synchronize the changed data from
B to A in preparation for returning to the local site.
Before doing this failback process, we need paths to be defined from B to A. For this task you
use the lsavailpprcport, mkpprcpath, and lspprcpath commands; see Example 24-36. You
give these commands to the DS HMC connected to the remote DS8000#2.
This process changes the A volume from its previous state (source) copy pending to target
copy pending; see Figure 24-9 on page 340. You have to add the -type gcp parameter with
the failbackpprc command to request Global Copy mode.
LSS10 LSS20
FlashCopy LSS22
A B C
GC: Target GC: Source FC: Target
Copy Pending Copy Pending
FC: Source
DS8000#1 DS8000#2
-dev IBM.2107-7520781 -dev IBM.2107-75ABTV1
Figure 24-9 Site swap scenario after Global Copy Failback from B to A
The volume at the production site becomes a write-inhibited target volume. This action is
performed on an individual volume basis.
Example 24-37 shows the commands given in our example. First we listed the status of the B
volumes and then performed the Global Copy Failback operation. The command is given to
the DS HMC connected to the remote DS8000#2.
340 IBM System Storage DS8000 Series: Copy Services in Open Environments
<< The failbackpprc B to A >>
dscli> failbackpprc -remotedev IBM.2107-7520781 -type gcp 2000-2001:1000-1001 2100-2101:1100-1101
Date/Time: November 12, 2005 12:04:37 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-75ABTV1
CMUC00197I failbackpprc: Remote Mirror and Copy pair 2000:1000 successfully failed back.
CMUC00197I failbackpprc: Remote Mirror and Copy pair 2001:1001 successfully failed back.
CMUC00197I failbackpprc: Remote Mirror and Copy pair 2100:1100 successfully failed back.
CMUC00197I failbackpprc: Remote Mirror and Copy pair 2101:1101 successfully failed back.
There is a -resetreserve parameter for the failbackpprc command. This option resets the
reserved state so that the failback operation can complete. In the failback operation after a
real disaster, you may use this parameter because the server might go down while the SCSI
persistent reserve was set on the A volume. If the server at the production site is operational,
for example, when a Global Mirror failback operation test is performed, you must not use this
parameter because the server at the local site still owns the A volume and may be using it and
the failback operation suddenly changes the contents of the volume. This may cause the
server file system’s corruption.
Example 24-38 Query for the Global Copy first pass completion
dscli> lspprc -l 2000-2001 2100-2101
Date/Time: November 12, 2005 12:06:11 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-75ABTV1
ID State Reason Type Out Of Sync Tracks Tgt Read Src Cascade Tgt Cascade Date Suspended SourceLSS Timeout (secs) Critical Mode First Pass
Status
===============================================================================================================================================================
2000:1000 Copy Pending - Global Copy 0 Disabled Disabled invalid - 20 unknown Disabled True
2001:1001 Copy Pending - Global Copy 0 Disabled Disabled invalid - 20 unknown Disabled True
2100:1100 Copy Pending - Global Copy 0 Disabled Disabled invalid - 21 unknown Disabled True
2101:1101 Copy Pending - Global Copy 0 Disabled Disabled invalid - 21 unknown Disabled True
dscli>
Depending on the host operating system, it may be necessary to dismount the B volumes.
Example 24-39 Query the Global Copy Out Of Sync Tracks until it shows zero
dscli> lspprc -l 2000-2001 2100-2101
Date/Time: November 12, 2005 12:06:11 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-75ABTV1
ID State Reason Type OutOfSyncTracks Tgt Read Src Cascade Tgt Cascade Date Suspended SourceLSS Timeout (secs) Critical Mode First Pass
================================================================================================================================================================
2000:1000 Copy Pending - Global Copy 0 Disabled Disabled invalid - 20 unknown Disabled True
2001:1001 Copy Pending - Global Copy 0 Disabled Disabled invalid - 20 unknown Disabled True
2100:1100 Copy Pending - Global Copy 0 Disabled Disabled invalid - 21 unknown Disabled True
2101:1101 Copy Pending - Global Copy 0 Disabled Disabled invalid - 21 unknown Disabled True
dscli>
In our example, the paths were still available; see Example 24-40. In there were no available
paths, then you must define them now using the mkpprcpath command; Example 24-1 on
page 307 shows how to do this task.
342 IBM System Storage DS8000 Series: Copy Services in Open Environments
IBM.2107-7520781/10 IBM.2107-75ABTV1/20 Success FF20 IBM.2107-7520781/I0213 IBM.2107-75ABTV1/I0140 5005076303FFC663
IBM.2107-7520781/11 IBM.2107-75ABTV1/21 Success FF21 IBM.2107-7520781/I0143 IBM.2107-75ABTV1/I0010 5005076303FFC663
IBM.2107-7520781/11 IBM.2107-75ABTV1/21 Success FF21 IBM.2107-7520781/I0213 IBM.2107-75ABTV1/I0140 5005076303FFC663
LSS10 LSS20
FlashCopy LSS22
A B C
GC: Source GC: Source FC: Target
Suspended Copy Pending
FC: Source
DS8000#1 DS8000#2
-dev IBM.2107-7520781 -dev IBM.2107-75ABTV1
Figure 24-10 Site swap scenario after Global Copy Failover from A to B
Example 24-41 shows the result of the failoverpprc command we used in our example, and
the volume state after this command is issued. You have to give this command to the DS
HMC connected to the local DS8000 (DS8000#1).
LSS10 LSS20
FlashCopy LSS22
A B C
GC: Source GC: Target FC: Target
Copy Pending Copy Pending
FC: Source
DS8000#1 DS8000#2
-dev IBM.2107-7520781 -dev IBM.2107-75ABTV1
Example 24-42 shows the result of the failbackpprc command used in our example, and the
volume state after this command is issued. You have to give this command to the DS HMC
connected to the local DS8000 (DS8000#1).
344 IBM System Storage DS8000 Series: Copy Services in Open Environments
CMUC00197I failbackpprc: Remote Mirror and Copy pair 1000:2000 successfully failed back.
CMUC00197I failbackpprc: Remote Mirror and Copy pair 1001:2001 successfully failed back.
CMUC00197I failbackpprc: Remote Mirror and Copy pair 1100:2100 successfully failed back.
CMUC00197I failbackpprc: Remote Mirror and Copy pair 1101:2101 successfully failed back.
dscli>
dscli> lspprc 1000-1001 1100-1101
Date/Time: November 12, 2005 12:15:36 AM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
==================================================================================================
1000:2000 Copy Pending - Global Copy 10 unknown Disabled True
1001:2001 Copy Pending - Global Copy 10 unknown Disabled True
1100:2100 Copy Pending - Global Copy 11 unknown Disabled True
1101:2101 Copy Pending - Global Copy 11 unknown Disabled True
LSS10 LSS20
FlashCopy LSS22
GM Master
Started 1000 2000 2200
Global Copy Pair
1001 2001 2201
LSS11 FlashCopy
LSS21 LSS23
A B C
GC: Source GC: Target FC: Target
Copy Pending Copy Pending
FC: Source
DS8000#1 DS8000#2
-dev IBM.2107-7520781 -dev IBM.2107-75ABTV1
The last step before starting the application at the production site is to start the Global Mirror
session again; see Figure 24-12, “Start Global Mirror” on page 345. If you did not already
create the FlashCopy relationships from B to C volumes, then you have to do it before starting
the Global Mirror.
Example 24-43 shows the commands used in our example and the corresponding results.
You run this command on the DS HMC connected to the local DS8000 (DS8000#1).
346 IBM System Storage DS8000 Series: Copy Services in Open Environments
24.5.10 Start the application at the local site
Now we have an environment on which to start the application at the original local site.
Depending on your operating system, it may be necessary to re-scan Fibre Channel devices
and mount the new source volumes (A volumes) at the local site. Start all applications and
check for consistency; see Figure 24-13. Depending on your path design, delete the paths
from the recovery to the production LSSs.
I/O
LSS10 LSS20
FlashCopy LSS22
GM Master
1000 2000 2200
Global Copy Pair
1001 2001 2201
LSS11 FlashCopy
LSS21 LSS23
A B C
GC: Source GC: Target FC: Target
Copy Pending Copy Pending
FC: Source
DS8000#1 DS8000#2
-dev IBM.2107-7520781 -dev IBM.2107-75ABTV1
Give this command to the DS HMC connected to the local disk subsystem.
348 IBM System Storage DS8000 Series: Copy Services in Open Environments
24.6.3 Pause Global Copy pairs
The pausepprc command suspends the Global Copy pairs; see Example 24-45. Give this
command to the DS HMC connected to the local disk subsystem.
Give this command to the DS HMC connected to the remote disk subsystem.
350 IBM System Storage DS8000 Series: Copy Services in Open Environments
2000:2200 20 0 300 Disabled Enabled Enabled Disabled Enabled Disabled Disabled
2001:2201 20 0 300 Disabled Enabled Enabled Disabled Enabled Disabled Disabled
2100:2300 21 0 300 Disabled Enabled Enabled Disabled Enabled Disabled Disabled
2101:2301 21 0 300 Disabled Enabled Enabled Disabled Enabled Disabled Disabled
FlashCopy
LSS10 LSS20 LSS22 LSS24
A B C D
GC: Source GC: Source FC: Target FC: Target
Copy Pending Suspended
FC: Source
DS8000#1 DS8000#2
-dev IBM.2107-7520781 -dev IBM.2107-75ABTV1
Example 24-50 shows the DS CLI log for this operation. We use the -nocp option for the
FlashCopy. You can also use the copy option.
FlashCopy
LSS10 LSS20 LSS22 LSS24
A B C D
GC: Source GC: Target FC: Target FC: Target
Copy Pending Copy Pending
FC: Source
DS8000#1 DS8000#2
-dev IBM.2107-7520781 -dev IBM.2107-75ABTV1
352 IBM System Storage DS8000 Series: Copy Services in Open Environments
For the failback operation we use the failbackpprc command; see Example 24-51.
Important: Do not specify the B volume as a source with the failbackpprc command (to
the DS8000#2), otherwise data on the B volume will be copied to the A volume. If the A
volume does not have reserve status, data on the A volume may be overwritten.
In our example, the first showgmir output shows Running in the Copy State field but the
Consistency Group formation is not done yet. You can find this from the Current Time and the
CG Time field. The next showgmir command shows that at least one Consistency Group
formation has been done after the showgmir command.
354 IBM System Storage DS8000 Series: Copy Services in Open Environments
CG Interval (seconds) 0
XDC Interval(milliseconds) 50
CG Drain Time (seconds) 30
Current Time 11/14/2005 21:31:15 JST
CG Time 11/14/2005 18:49:59 JST
Successful CG Percentage 99
FlashCopy Sequence Number 0x43785DC7
Master ID IBM.2107-7520781
Subordinate Count 0
Master/Subordinate Assoc -
dscli> showgmir -dev IBM.2107-7520781 IBM.2107-7520781/10
Date/Time: November 14, 2005 9:26:03 PM JST IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520781
ID IBM.2107-7520781/10
Master Count 1
Master Session ID 0x02
Copy State Running
Fatal Reason Not Fatal
CG Interval (seconds) 0
XDC Interval(milliseconds) 50
CG Drain Time (seconds) 30
Current Time 11/14/2005 21:31:30 JST
CG Time 11/14/2005 21:31:30 JST
Successful CG Percentage 99
FlashCopy Sequence Number 0x437883A2
Master ID IBM.2107-7520781
Subordinate Count 0
Master/Subordinate Assoc -
In the DS SM GUI examples we use two DS8000s, serial numbers 7503461 and 75ABTV1, in
a quite similar configuration as the one used in the previous examples. Still, in this case note
that 7503461 is at the local production site and 75ABTV1 is at the remote backup site. Also,
on each machine we use volumes in LSS 14. You can tell the volumes apart by their names.
The local volumes on serial 7503461 are named av_8K_1_1400 to av_8K_1_1403. The
remote volumes on serial 75ABTV1 are named av_8K_3_1400 to av_8K_3_1403.
We had Ethernet connectivity between the sites to connect to the storage complexes. We also
performed the procedure to add the remote storage complex to the local storage complex.
To launch this wizard you need to go first to the Paths panel under the Copy Services menu of
the DS Storage Manager GUI; see Figure 24-16. Select from the pull-down lists the storage
complex, then the storage unit, then the storage image, and finally the LSS that contains the
source volumes of the Global Copy pairs we want to create. In the Select Action pull-down
list, choose Create to proceed with the first step of the wizard.
Figure 24-16 Global Copy paths creation - Launch the creation process
Now the creation wizard is displaying the Select source LSS panel; see Figure 24-17. Here
you select from the pull-down list, the LSS that contains the source volumes of the Global
Copy pairs you are going to create.
Figure 24-17 Global Copy paths creation - step 1: select the source LSS
356 IBM System Storage DS8000 Series: Copy Services in Open Environments
When the creation wizard displays the Select target LSS panel, select from the pull-down lists
the storage complex, then the storage unit, then the storage image, and finally the LSS that
contains the target volumes; see Figure 24-18.
Figure 24-18 Global Copy paths creation - step 2: select the target LSS
When the creation wizard displays the Select source I/Os ports panel, select (using the check
boxes) at least one I/O port (two is better) to use for Global Copy replication; see
Figure 24-19. In the location column, four digits indicate the location of a port:
The first digit (R) is to locate the frame.
The second digit (E) is for the I/O enclosure.
The third digit (C) is for the adapter.
The fourth digit (P) is for the adapter’s port.
Figure 24-19 Global Copy paths creation - step 3: select the source I/O ports
In this example, no choices are being presented, as there is only one path for our port.
Figure 24-20 Global Copy paths creation - step 4: select the target I/O ports
When the creation wizard displays the Select path options panel, you can select using the
check box the option Define as Consistency Group; see Figure 24-21. This option is not
mandatory since the Global Mirror session will handle the consistency of the data across the
set of volumes.
Figure 24-21 Global Copy paths creation - step 5: select the paths options
Click Next to proceed with the sixth and last step of this wizard.
358 IBM System Storage DS8000 Series: Copy Services in Open Environments
Figure 24-22 shows the Verification panel. Here you check all the components of your paths
configuration and, if necessary, click Back to correct any of them or click Finish to validate
the configuration and end the wizard.
Note: If Global Copy pairs are in several LSSs, do not forget to select all of them during
this wizard or you will have to run the wizard again on each LSS. If Global Copy pairs are
spread over several storage units, run this wizard again on each of them.
To launch this wizard, you need to go first to the Metro Mirror panel under the Copy Services
menu of the DS Storage Manager GUI; see Figure 24-23. In the Select Action pull-down list,
choose Create.
Figure 24-24 Global Copy creation - step 1: choose volume pairing method
When the creation wizard displays the Select source volumes panel, select from the
pull-down lists the storage complex, the storage unit, the storage image, the resource type,
and if necessary its appropriate parameter to display the list of volumes. See Figure 24-25.
Figure 24-25 Global Copy creation - step 2: select the source volumes
If you have chosen the resource type LSS, then select from the pull-down list the LSS number
that contains the source volumes you are going to use for Global Mirror and then check the
boxes of the selected volumes.
360 IBM System Storage DS8000 Series: Copy Services in Open Environments
Note: At this step, if paths have not yet been created we can click Create Paths to launch
the wizard described in 24.8.1, “Define paths” on page 355.
When the creation wizard displays the Select target volumes panel, we can notice that only
one source volume is indicated on the top of the panel; see Figure 24-26. This means that we
will repeat this third step for each volume that we have selected during the second step.
Figure 24-26 Global Copy creation - Step 3: select the target volume 1
Select from the pull-down lists the storage complex, then the storage unit, then the storage
image, then the resource type, and if it is necessary its appropriate parameter to display the
list of volumes. Then on the required LSS check the box for the selected target volume.
Click Next to proceed with the selection of the second target volume; see Figure 24-27.
Figure 24-27 Global Copy creation - Step 3: select the target volume 2
If we had selected more source volumes, we would once again proceed to the next target
volume selection panel. Since it is the second and last volume in our selection, click Next to
proceed with the fourth step of this wizard.
When the creation wizard displays the Select copy options panel, select the radio button
Global Copy to define the type of replication, then check the box for Permit read access
from target, and if it is the first synchronization between source and target volumes, check
the box for Perform initial copy. See Figure 24-28.
Figure 24-28 Global Copy creation - Step 4: select the copy options
Click Next to proceed with the fifth and last step of this wizard.
362 IBM System Storage DS8000 Series: Copy Services in Open Environments
In the Verification panel, verify all the components of your Global Copy session configuration
and, if necessary, click Back to correct any of them or click Finish to validate; see
Figure 24-29.
Note: If FlashCopy pairs are in several LSSs, do not forget to select all of them during this
wizard or you will have to run the wizard again on each LSS. If FlashCopy pairs are spread
over several storage images run this wizard again on each of them.
Because they are for a Global Mirror environment, these FlashCopy pairs are created on the
remote machine using the Global Copy targets as the FlashCopy source. This means we
need to select the remote storage complex and the remote storage unit.
When the creation wizard displays the Define relationship type, select the radio button A
single source with a single target. This is because we will have to activate the record option
on the fourth step of this wizard; see Figure 24-31.
364 IBM System Storage DS8000 Series: Copy Services in Open Environments
When the creation wizard displays the Select source volumes panel, select from the
pull-down lists the storage complex, then the storage unit, then the storage image, then the
resource type, and if it is necessary its appropriate parameter to display the list of volumes. If
you have chosen the resource type LSS, select from the pull-down lists the LSS number that
contains the source volumes you want to use for the Global Mirror environment. Then check
the boxes of the selected source volumes. See Figure 24-32.
When the creation wizard displays the Select target volumes panel, select from the pull-down
lists the resource type and, if it is necessary, its appropriate parameter to display the list of
volumes. If you have chosen the resource type LSS, select from the pull-down lists the LSS
number that contains the target volumes you want to use. Then check the boxes of the
selected target volumes. See Figure 24-33.
Note: Do not check the box for Initiate background copy. You may need to un-select it.
Click Next to proceed with the fifth and last step of this wizard.
In the Verification panel, verify all the components of the FlashCopy configurations, and if
necessary click Back to correct any of them or click Finish to validate. See Figure 24-35.
366 IBM System Storage DS8000 Series: Copy Services in Open Environments
24.8.4 Create the Global Mirror session
With the DS Storage Manager, to create a Global Mirror session you go through a three-step
process.
To launch this wizard you need to go first to the Global Mirror panel under the Copy Services
menu of DS Storage Manager GUI; see Figure 24-36. Select from the pull-down lists the
storage complex, then the storage unit, and then the storage image that contains the source
volumes that will be part of the Global Mirror. Then from the Select Action pull-down list,
choose Create.
The creation wizard displays the Select volumes panel; see Figure 24-37.
Note: At this step, if Global Copy pairs and FlashCopy pairs are not yet created, we can
click Create Metro Mirror to launch the wizard described in 24.8.2, “Create Global Copy
pairs” on page 359, and we can click Create FlashCopy to launch the wizard described in
24.8.3, “Create FlashCopy relationships” on page 363.
When the creation wizard displays the Define properties panel, the session number field
should be filled with the appropriate session number. We can click Get Available Session
Numbers if we have forgotten the one we want to use. See Figure 24-38.
There are also three important fields that can be set at this moment using this panel, and
these are the Global Mirror session tuning parameters. For a detailed understanding of these
options, see 20.4.2, “Consistency Group parameters” on page 261. Following is a brief
discussion of each of them:
Consistency Group interval time (seconds): specifies how long to wait between the
formation of Consistency Groups. If this number is not specified or is set to zero,
Consistency Groups are formed continuously.
Maximum coordination interval (milliseconds): indicates the maximum time that Global
Mirror can queue I/Os in the source disk subsystem to start forming a Consistency Group.
Maximum time writes inhibited to remote site (seconds): specifies the maximum amount of
time allowed for the consistent set of data to drain to the remote site before failing the
current Consistency Group.
Click Next to proceed with the third and last step of this wizard.
368 IBM System Storage DS8000 Series: Copy Services in Open Environments
In the Verification panel, check all the components of your Global Mirror session configuration
and, if necessary, click Back to correct any of them or click Finish to validate; see
Figure 24-39.
To view the status of the Global Mirror session, go to the Global Mirror panel under the Copy
Services menu of the DS Storage Manager GUI. Select from the pull-down lists the storage
complex, then the storage unit, then the storage image, which is the master Global Mirror
session manager, and wait until the screen is refreshed. If it is necessary, click Refresh to
refresh the panel. See Figure 24-40.
24.9.1 View settings and error information of the Global Mirror session
To see session information you first to go to the Global Mirror panel under the Copy Services
menu of the DS Storage Manager GUI. Then check the box for the Global Mirror session for
which you want to display its properties, and in the Select Action pull-down list, choose
Properties; see Figure 24-41.
Figure 24-41 View Global Mirror session’s properties: Launch the viewing process
The Global Mirror session properties: Real-time panel will be displayed. In the General tab
you can review the setting for this session; see Figure 24-42.
370 IBM System Storage DS8000 Series: Copy Services in Open Environments
You can then click Failures to see the Failures tab; see Figure 24-43. In the Global Mirror
session properties: Real-time panel, on the Failures tab, you can request selected information
using the radio buttons Most recent failure, Previous failure, and First failure. Once you are
done reviewing the information, click OK to finish and go back to the main Global Mirror
panel.
Figure 24-43 View Global Mirror session errors information - Failures tab
The Global Mirror session volumes: Real-time panel will be displayed and it will show
information about the volumes associated with the Global Mirror session; see Figure 24-44.
You can either download or print this information table. Then click OK to finish and go back to
the main Global Mirror panel.
When the Global Mirror: Real-time panel displays the warning message (see Figure 24-45)
either click Cancel to return to the main Global Mirror panel without pausing the Global Mirror
session or OK to pause the Global Mirror session and to return to the main Global Mirror
panel.
Figure 24-45 Pause Global Mirror: Confirm the pause of the Global Mirror session
When the main Global Mirror panel is displayed, note that the state of the Global Mirror
session is now Paused; see Figure 24-46.
372 IBM System Storage DS8000 Series: Copy Services in Open Environments
When the Global Mirror Real-time panel displays the warning message (see Figure 24-47),
either click Cancel to return to the main Global Mirror panel without resuming the Global
Mirror session or click OK to resume the Global Mirror session and to return to the main
Global Mirror panel. When the main Global Mirror panel is displayed, the state of the Global
Mirror session will now show Running.
Figure 24-47 Resume Global Mirror - Confirm the resume of the Global Mirror session
To launch the wizard you first go to the Global Mirror panel under the Copy Services menu of
the DS Storage Manager GUI. Then select the Global Mirror session you wish to work with.
Then in the Select Action pull-down list, choose Modify to proceed with the first step of the
wizard; see Figure 24-41 on page 370.
When the modification wizard displays the Select volumes panel, click the required storage
unit, then the required LSS, and either select or un-select the check box for the source
volumes you wish to either add or remove from the Global Mirror session. The panel will look
similar to Figure 24-37 on page 367.
Note: At this step, if Global Copy pairs and FlashCopy pairs have not been created yet,
you can click Create Metro Mirror to launch the wizard described in 24.8.2, “Create Global
Copy pairs” on page 359, and you can click Create FlashCopy to launch the wizard
described in 24.8.3, “Create FlashCopy relationships” on page 363.
When the creation wizard displays the Define properties panel, the session number field
should be filled with the appropriate session number. This panel will look similar to
Figure 24-38 on page 368. In this panel there are three important fields you can modify at this
moment: Consistency Group interval time (seconds), Maximum coordination interval
(milliseconds), and Maximum time writes inhibited to remote site (seconds). For a detailed
discussion of these Global Mirror tuning parameters see 20.4.2, “Consistency Group
parameters” on page 261.
Click Next to proceed with the third and last step of this wizard. This is the Verification panel.
Here you verify all the components of your Global Mirror session configuration and, if
necessary, click Back to correct any of them or click Finish to validate.
Part 7 Metro/Global
Mirror
This part gives an understanding of a new function for the DS8000, Metro/Global Mirror, a
3-site disaster recovery solution.
Server or Servers
***
normal application
I/Os Global Mirror network Global Mirror FlashCopy
asynchronous incremental
Metro Mirror long distance NOCOPY
A B C
D
Metro Mirror
network Global Mirror
synchronous
short distance
Both Metro Mirror and Global Mirror are well established replication solutions. Metro/Global
Mirror combines Metro Mirror and Global Mirror to incorporate the best features of the two
solutions:
Metro Mirror
– Synchronous operation supports zero data loss.
– The opportunity to locate the intermediate site disk subsystems close to the local site
disk subsystems allows use of intermediate site disk subsystems in a high availability
configuration.
Global Mirror
– Asynchronous operation supports long distance replication for disaster recovery.
– Global Mirror methodology has no impact on applications at the local site.
– Provides a recoverable, restartable, consistent image at the remote site with a
Recovery Point Objective (RPO) typically in the 3–5 second range.
378 IBM System Storage DS8000 Series: Copy Services in Open Environments
This chapter provides a high level overview of Metro/Global Mirror. The details of the
individual processes/elements of the solution, for example, Metro Mirror and Global Mirror,
will not be repeated here as they are already described in great detail in respective parts of
this book.
The local site (site A) to intermediate site (site B) component is identical to Metro Mirror.
Application writes are synchronously copied to the intermediate site before write complete is
signaled to the application. All writes to the local site volumes in the mirror are treated in
exactly the same way. This is explained in great detail in Chapter 11, “Metro Mirror overview”
on page 109, of this book.
The intermediate site disk subsystems are collectively paused by the Global Mirror master
disk subsystem to create the Consistency Group (CG) set of updates. This pause would
normally take 3 ms every 3 to 5 seconds. After the CG set is formed, the Metro Mirror writes,
from local site (site A) volumes to intermediate site (site B) volumes, are allowed to continue.
Also, the CG updates continue to drain to remote site (site C) volumes. The intermediate site
to remote site drain is expected to take only a few seconds to complete, perhaps as few as 2
or 3 seconds.
Once all updates are drained to the remote site, all changes since the last FlashCopy from
the C volumes to the D volumes are logically (NOCOPY) FlashCopied to the D volumes. After
the logical FlashCopy is complete, the intermediate site to remote site Global Copy data
transfer is resumed until the next formation of a Global Mirror Consistency Group. The
process described above is repeated every 3 to 5 seconds if the interval for Consistency
Group formation is set to zero. Otherwise it will be repeated at the specified interval plus 3 to
5 seconds.
The Global Mirror processes are presented in much greater detail in Chapter 20, “Global
Mirror overview” on page 243.
Server or Servers
***
4
normal application I/Os
Global Mirror network Global Mirror FlashCopy
asynchronous incremental
Metro Mirror long distance NOCOPY
1
2
A B C
3 b
a c
D
Metro Mirror network
synchronous Global Mirror
short distance
380 IBM System Storage DS8000 Series: Copy Services in Open Environments
26
One method to be prepared for a disaster at the remote site is to use the storage at the
intermediate site for a target copy. Because the bandwidth between the remote and the
intermediate site will generally be too low to set up a Metro Mirror, the only possible way to
copy the data is to establish a Global Copy relationship. Global Copy by itself does not
guarantee that the copied data is consistent, but a Global Mirror relationship can provide the
necessary data consistency. To set up a Global Mirror between the remote and intermediate
sites, an additional set of FlashCopy volumes is required at the intermediate site. Figure 26-1
illustrates the volume setup for Metro/Global Mirror with an additional Global Mirror.
Global Mirror
E from remote
to intermediate site
A B C
Metro Mirror
Global Mirror
D
from intermediate
to remote site
Figure 26-1 Metro/Global Mirror with additional Global Mirror from remote to intermediate site
382 IBM System Storage DS8000 Series: Copy Services in Open Environments
26.1.2 Metro/Global Mirror with multiple storage subsystems
Based on the capabilities of Global Mirror, it is possible to span the Metro/Global Mirror
across multiple storage subsystems. At the intermediate site one of the storage subsystems
has the role of the Global Mirror master and the other ones are the subordinates. The Global
Mirror master controls the formation of the Consistency Group. It requires additional RMC
(PPRC) links between the master and each subordinate. See Figure 26-2.
The Metro Mirror will be set up from multiple storage subsystems at the local site to the
intermediate to the Global Mirror primary systems. The paths of the Metro Mirror relations
must be established with consistency group enabled.
Global Mirror
Master
Metro Mirror
Global Mirror
Global Mirror
Metro Mirror
Subordinate
Global Mirror
Metro Mirror
Subordinate
In a disaster situation where a failover to the intermediate or remote site has to be done, all
servers and storage that are required to start up an application are subject to the takeover.
Thus a disaster recovery concept must be worked out for each application individually. It
defines which components and functionality must be recovered at the intermediate or remote
sites in case of a disaster.
Note: The schematic in Figure 26-3 shows only one Fibre Channel director per site. In a
real implementation the connections between the sites should be realized by two
redundant fabrics across all locations.
Primary Secondary
production site production site Remote site
Global Mirror
Metro Mirror
Global Mirror
Metro Mirror
2109-M14
2109-M14
P590
P590
HACMP Remote
stretched clusters production for
stretched clusters
When an application uses a Metro Mirror from the primary production site to the secondary
production site, the primary production site is the Metro/Global Mirror local site and the
secondary site is the intermediate site and vice versa for applications running in the other
direction.
384 IBM System Storage DS8000 Series: Copy Services in Open Environments
This setup allows you to configure cluster systems that span across both production sites,
offering more flexibility for high availability setups. A failure of a server at the production site
can be taken over automatically at the other production site using the automatic take over
feature of the cluster software while the primary storage is still being used. Otherwise, if a
single host system, as shown in Figure 26-3 on page 384, fails at the primary production site,
the only possibility to bring up the production again is a recovery of the storage and the server
at the remote site.
A failure of the storage can be seen as a failure of an infrastructure component, and this can
be categorized as at least a partial disaster. In this case, a recovery of storage at the
intermediate or remote site has to be performed. The storage at the intermediate site can be
accessed by the server at the local site, when the bandwidth between both production sites is
high enough. This implies that there is not necessarily a takeover of the servers required,
which offers more flexibility on how to start up the applications.
For cost effectiveness it is possible to consolidate the needed disk capacity at the remote site.
Because of the distance, a stretched cluster environment may not be possible, so single host
systems or local clustered systems are implemented at the remote site. A large scale server
which is capable of providing multiple logical partitions to run the multiple applications from
the production sites could be used. It is also possible to equip the storage subsystems at the
remote site with disk drive modules of higher capacity to reduce the number of installed
storage subsystems.
Figure 26-4 shows the logical setup of the Metro/Global Mirror. It also shows the steps in the
order of the operational sequence. Considering the potentially large amount of data that has
to be copied, we recommend that you first set up Global Copy with the NOCOPY option.
When the Metro Mirror is created and starts to synchronize from the local site to the
intermediate site, the Global Copy passes the tracks to the remote site.
To avoid performance impact on a running production, we recommend that you start the
Metro Mirror initial copy when the production write activity is low.
A 1 3 B 1 2 C
4
6
D
5
To identify the ports to which the PPRC links are connected you can use the lsavailpprcport
command. This command displays all local and remote ports which have a physical
connection. This could be either a point-to-point link or a zoned connection through a storage
area network.
In Example 26-1 you can see that there are four PPRC links (physical paths) available. These
connections can be used to create the PPRC paths according to the user’s installation plan.
26.3.2 Step 1: Set up all Metro Mirror and Global Mirror paths
Based on the PPRC links that are discovered, the paths have to be created for each LSS pair
in the Metro/Global Mirror configuration. There are two prerequisites which have to be fulfilled
to create PPRC paths successfully:
The physical connection must be available, either by an appropriate zone in a storage
area network or by point-to-point links.
The LSSs must exist in both storage devices. This means that at least one volume of the
designated LSSs must exist.
It is a good practice to create the PPRC paths for both directions. In case of a disaster all
paths in all directions are available and do not need to be created in the critical situation of the
disaster.
386 IBM System Storage DS8000 Series: Copy Services in Open Environments
Example 26-2 shows the setup of Metro Mirror paths between the local and intermediate
sites. The mkpprcpath command is always executed at the DS8000 where the source
volumes are for each direction. This means that for the direction local to intermediate, the
mkpprcpath is executed at the local site and vice versa for the opposite direction.
Example 26-2 Set up PPRC paths between local and intermediate site
At local site:
At intermediate site:
Note: To ensure consistency of the Metro Mirror, the PPRC paths have to be created with
the option -consistgrp. Between the intermediate site and the remote site, the Global Mirror
is responsible for the creation of the Consistency Group. This means that the paths for the
Global Copy must not be set up with the -consistgrp option.
Example 26-3 shows the mkpprcpath command for the Global Mirror. Again, the paths are
created for both directions and the commands are executed at the DS8000 where the source
is.
Example 26-3 Set up PPRC paths between intermediate site and remote site
At intermediate site:
At remote site:
26.3.4 Step 3: Set up Metro Mirror between local and intermediate sites
To set up the Metro Mirror the mkpprc command is used, where the option -type mmir signifies
that a Metro Mirror will be created. For the initial setup the pairs are created in the full copy
mode. Example 26-5 shows the mkpprc command, which is executed at the local site.
Example 26-5 Set up Metro Mirror from the local site to the intermediate sites
dscli> mkpprc -dev IBM.2107-75ABTV1 -remotedev IBM.2107-75ABTV2 -type mmir -mode full
7000-7003:7200-7203
Date/Time: June 19, 2006 3:58:51 PM CEST IBM DSCLI Version: 5.1.600.196 DS: IBM.2107-75ABTV1
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 7000:7200 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 7001:7201 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 7002:7202 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 7003:7203 successfully created.
dscli>
The volumes at the intermediate site are target volumes for Metro Mirror and source volumes
for Global Copy at the same time. When the lspprc command is executed against these
volumes, it shows the pair status of both the Metro Mirror and Global Copy relationships, as
illustrated in Example 26-6.
388 IBM System Storage DS8000 Series: Copy Services in Open Environments
IBM.2107-75ABTV2/7200:IBM.2107-75BYGT1/7400 Copy Pending - Global Copy
IBM.2107-75ABTV2/72 unknown Disabled True
IBM.2107-75ABTV2/7201:IBM.2107-75BYGT1/7401 Copy Pending - Global Copy
IBM.2107-75ABTV2/72 unknown Disabled True
IBM.2107-75ABTV2/7202:IBM.2107-75BYGT1/7402 Copy Pending - Global Copy
IBM.2107-75ABTV2/72 unknown Disabled True
IBM.2107-75ABTV2/7203:IBM.2107-75BYGT1/7403 Copy Pending - Global Copy
IBM.2107-75ABTV2/72 unknown Disabled
True
Example 26-7 shows how to create the FlashCopy for the Global Mirror at the remote site.
Example 26-7 Create the FlashCopy for the Global Mirror at the remote site
dscli> mkflash -dev IBM.2107-75BYGT1 -tgtinhibit -record -persist -nocp 7400-7403:7600-7603
Date/Time: June 19, 2006 4:04:17 PM CEST IBM DSCLI Version: 5.1.600.196 DS:
IBM.2107-75BYGT1
CMUC00137I mkflash: FlashCopy pair 7400:7600 successfully created.
CMUC00137I mkflash: FlashCopy pair 7401:7601 successfully created.
CMUC00137I mkflash: FlashCopy pair 7402:7602 successfully created.
CMUC00137I mkflash: FlashCopy pair 7403:7603 successfully created.
26.3.6 Step 5: Create Global Mirror session and add volumes to session
This is done at the intermediate site. The session is created with the mksession command.
For each LSS a session has to be created. The session is denoted by a session number. Only
one session can be active in a storage subsystem. This means that all LSSs in a Global
Mirror must have the same session number.
390 IBM System Storage DS8000 Series: Copy Services in Open Environments
With the command showgmir it can be verified if the Global Mirror was successfully created.
The Copy State should show Running. See Example 26-10.
When the option -metrics is supplied with the showgmir command the progress of the
Consistency Group formation can be monitored. The entry Total Successful CG Count shows
the current number of successful created Consistency Groups. When the Global Mirror is
running the number of Consistency Groups is steadily growing each time the showgmir
command is issued. See Example 26-11.
The command showgmiroos displays the number of tracks which are out of synchronization.
Example 26-12 shows the OutOfSyncTracks of LSS 72.
The status of the session is displayed with the command lssession. It displays the session
status of all volumes of an LSS. The command can take a list of LSSs to inspect the session
status of volumes from multiple LSSs.
In this section we describe the procedure on how to migrate from a two-site Metro Mirror
setup (local site to intermediate site) to a three-site Metro/Global Mirror setup (local site to
intermediate site to remote site) using DS CLI. This can be done while production is up and
running at the local site. Depending on the capability of the link between the intermediate site
and the remote site, the initial copy of the data may be more or less influenced by write
activity of the applications. If possible, the initial copy should be started during a low write
392 IBM System Storage DS8000 Series: Copy Services in Open Environments
activity of the application. But in any case, the first pass of the Global Copy must be
completed for the Global Mirror to be able to form consistency groups.
Note: To ensure consistency from the local to the remote site it is essential that the Metro
Mirror paths are created with consistency enabled. If this was not the case for the existing
Metro Mirror, the paths for each LSS can be changed using the chlss command.
A B 1 2 C
3
4
D
5
As shown in Figure 26-5, the steps required to extend an existing Metro Mirror to a three-site
setup using Metro/Global Mirror are:
1. Set up PPRC paths from the intermediate site to the remote site.
2. Set up Global Copy with COPY from the intermediate site to the remote site.
In contrast to the Metro/Global Mirror setup, the Global Copy is established with copy
mode to ensure that all data is copied from the intermediate site to the remote site.
3. Set up FlashCopy at the remote site.
4. Create a Global Mirror session and add volumes to the session at the intermediate site.
5. Start Global Mirror at the intermediate site.
The steps are described in detail in 26.3, “Initial setup of Metro/Global Mirror” on page 385.
394 IBM System Storage DS8000 Series: Copy Services in Open Environments
27
396 IBM System Storage DS8000 Series: Copy Services in Open Environments
In Example 27-1 all volumes of the Metro Mirror are in Full Duplex mode.
The Global Copy volume status will always be Copy Pending. To identify that all tracks were
copied to the secondary site the Out of Sync Tracks must be obtained. The Global Copy is
synchronized when for all volume pairs the out Out of Sync Tracks has the value zero, as
shown in Example 27-2.
In a Metro/Global Mirror the volumes at the intermediate site are in a different state than in a
conventional Metro Mirror or Global Mirror. Because the Global Mirror is cascaded to the
Metro Mirror, the volumes at the intermediate site are target and source at the same time.
Thus the lspprc command issued at the intermediate site will show both the Metro Mirror and
the Global Mirror, as shown in Example 27-3.
For a detailed description of how consistency is provided in a Metro Mirror see 12.4,
“Consistency Group function” on page 116.
As shown in Example 27-4, the pair status at the primary site after the freezepprc is now
Suspended with the reason Freeze. At the secondary site the pairs status is still in Target Full
Duplex mode.
Example 27-4 Example of a freezepprc and the pair status at the primary site
At the local DS8000:
398 IBM System Storage DS8000 Series: Copy Services in Open Environments
dscli> lspprc -dev IBM.2107-75ABTV1 -remotedev IBM.2107-7503461 -fmt default 6200-6203
ID State Reason Type SourceLSS Timeout (secs) Critical Mode
First Pass Status
===========================================================================================
==============
6800:6200 Target Full Duplex - Metro Mirror 68 unknown Disabled
Invalid
6801:6201 Target Full Duplex - Metro Mirror 68 unknown Disabled
Invalid
6802:6202 Target Full Duplex - Metro Mirror 68 unknown Disabled
Invalid
6803:6203 Target Full Duplex - Metro Mirror 68 unknown Disabled
Invalid
The unfreezepprc removes the long busy status from the primary volumes and I/O will
continue. The pair status of the primary volumes is still Suspended, as shown in Example 27-5
Important: We recommend that you run both commands in one script to ensure that the
time delay between the freeze and the unfreeze is as small as possible.
To achieve this, the failed-over volumes have to be removed from the Global Mirror session.
This removes these volumes from the focus of the Global Mirror, and the other volumes are
processed by the Global Mirror. To ensure that the volumes that have to be removed from the
session are consistent, the Global Mirror should be paused until all volumes are removed.
After the Global Mirror is resumed, Consistency Groups will be formed for the remaining
volumes and the failover procedure can continue. Example 27-6 shows the commands.
The Global Mirror is a process running at the intermediate site, which uses the FlashCopy at
the remote site. When more than one storage subsystem is used at the intermediate site, one
of these is denoted as the master where the Global Mirror resides (see 26.1.2, “Metro/Global
Mirror with multiple storage subsystems” on page 383). The other storage subsystems are
the subordinates. The Global Mirror coordinates the consistency formation for all volumes of
400 IBM System Storage DS8000 Series: Copy Services in Open Environments
the master and the subordinates, which are joined into the Global Mirror session. The
consistency is formed in three steps:
1. The master is coordinating with all subordinates to stop the I/O for 3–5 msec to all primary
volumes to form a Consistency Group.
2. The Consistency Group is transmitted to the secondary volumes.
3. The FlashCopy copies all tracks that have been changed since the last write operation to
the secondary volumes.
See 20.4, “Consistency Groups” on page 259, for a detailed description of cosistency
formation in a Global Mirror.
Drain consistency group and send to FlashCopy committed Start next consistency group
remote DS8000 using Global Copy once all revertible
Application writes for next flashcopies have
consistency group are recorded in successfully completed
change recording bitmap
Maximum drain time - eg 1 min
The consistency of the data can be provided in each phase of the consistency group
formation process, as shown in Figure 27-1:
When the failure occurs during the coordination time or the draining of the data to the
remote site, consistency is still available on the FlashCopy volumes, because the
FlashCopy has not yet started.
When the failure happens while the FlashCopy command is executed, a manual
intervention is required to either roll revert or commit the consistency group before
continuing with the recovery or restarting Global Mirror. The action that is indicated
depends on the current status of the FlashCopy, where the sequence numbers and the
revertible flag are of special interest:
– When the sequence numbers of the FlashCopy are different, the copy process has not
started for all the volumes. In this case the recent FlashCopy is inconsistent and can
not be used and must be rolled back with the revertflash command, which removes
all not-committed sequences from the FlashCopy target.
– When the sequence numbers are all equal and there is a mix of revertible and
unrevertible volumes, the copy to the FlashCopy targets has taken place but the
process has not been finished for some volumes. In this case the resent FlashCopy
targets are usable and the process has to be committed manually by a commitflash.
.....
50E7:5AE7 50 437DABED 300 Disabled Enabled Enabled Disabled Enabled
Disabled Disabled 15259 Fri Nov 18 09:22:09 CET 2005 Fri Nov 18
11:24:44 CET 2005
50E8:5AE8 50 437DABED 300 Disabled Enabled Enabled Enabled Disabled
Disabled Disabled 15259 Fri Nov 18 09:22:09 CET 2005 Fri Nov 18
11:24:44 CET 2005
50E9:5AE9 50 437DABED 300 Disabled Enabled Enabled Enabled Disabled
Disabled Disabled 15259 Fri Nov 18 09:22:09 CET 2005 Fri Nov 18
11:24:44 CET 2005
.....
Example 27-8 shows a revertible situation. The sequence numbers have two different values,
whereby for the lower sequence number the revertible flag is disabled. This shows that for
these volumes the FlashCopy process has not yet started. The volumes with the higher
sequence numbers have the revertible flag enabled, which means that the relationship exists
but is not committed. The correct interaction to bring back consistency is to issue a
revertflash command.
.....
50E3:5AE3 50 437DC7BD 300 Disabled Enabled Enabled Enabled Disabled
Disabled Disabled 15259 Fri Nov 18 09:22:09 CET 2005 Fri Nov 18
13:23:24 CET 2005
402 IBM System Storage DS8000 Series: Copy Services in Open Environments
50E4:5AE4 50 437DC7BC 300 Disabled Enabled Enabled Disabled Enabled
Disabled Disabled 15259 Fri Nov 18 09:22:09 CET 2005 Fri Nov 18
13:23:23 CET 2005
50E5:5AE5 50 437DC7BD 300 Disabled Enabled Enabled Enabled Disabled
Disabled Disabled 15259 Fri Nov 18 09:22:09 CET 2005 Fri Nov 18
13:23:24 CET 2005
50E6:5AE6 50 437DC7BC 300 Disabled Enabled Enabled Disabled Enabled
Disabled Disabled 15259 Fri Nov 18 09:22:09 CET 2005 Fri Nov 18
13:23:23 CET 2005
50E7:5AE7 50 437DC7BD 300 Disabled Enabled Enabled Enabled Disabled
Disabled Disabled 15259 Fri Nov 18 09:22:09 CET 2005 Fri Nov 18
13:23:24 CET 2005
50E8:5AE8 50 437DC7BC 300 Disabled Enabled Enabled Disabled Enabled
Disabled Disabled 15259 Fri Nov 18 09:22:09 CET 2005 Fri Nov 18
13:23:23 CET 2005
50E9:5AE9 50 437DC7BC 300 Disabled Enabled Enabled Disabled Enabled
Disabled Disabled 15259 Fri Nov 18 09:22:09 CET 2005 Fri Nov 18
13:23:23 CET 2005
50EA:5AEA 50 437DC7BD 300 Disabled Enabled Enabled Enabled Disabled
Disabled Disabled 15259 Fri Nov 18 09:22:09 CET 2005 Fri Nov 18
13:23:24 CET 2005
50EB:5AEB 50 437DC7BD 300 Disabled Enabled Enabled Enabled Disabled
Disabled Disabled 15259 Fri Nov 18 09:22:09 CET 2005 Fri Nov 18
13:23:24 CET 2005
When the option -revertible is supplied with the lsflash command, only the revertible
volumes are listed.
As a final step to provide complete and consistent data, after a failover to a remote site, the
data located at the FlashCopy target volumes must be reversed to the source volumes. After
the failover to the Global Mirror secondary site, the host access is gained to the secondary
volumes, which are also the FlashCopy source volumes.
After a failover to the remote site the host access is gained to the secondary volumes of the
Global Copy relation. But some last-saved consistent data may be located at the FlashCopy
target volumes. In this case the FlashCopy target volumes must be reversed to their sources,
which are also the secondary Global Copy volumes.
Note: When the application has been stopped and two consistency groups have been
formed, we can assume that the data on the FlashCopy source and on the FlashCopy
target are the same. In this case a fast reverse restore is not necessary.
To do so requires that additional volumes are available to set up a FlashCopy relation with the
volumes at the intermediate site as the FlashCopy source volumes. After the failover to the
remote site is completed (see 28.4, “Recovery at remote site” on page 416, for the details) a
failback of the Global Copy will be used to copy the data to the intermediate site. Finally, a
session with the related volumes must be created and the Global Mirror must be started at
the remote site.
Figure 27-2 shows the steps to set up the additional Global Mirror.
E
2
A B C
3
Figure 27-2 Set up an additional Global Mirror after failover to remote site
The setup of the additional Global Mirror comprises the following steps:
1. Create or fail back from the remote site to the intermediate site.
2. Establish FlashCopy to the additional volumes at the intermediate site.
3. Create a session and start up the Global Mirror.
404 IBM System Storage DS8000 Series: Copy Services in Open Environments
Example 27-10 shows how to set up the Global Copy. Because the data at the remote site is
the same as at the intermediate site, the Global Copy can be established with the option
-mode nocp to avoid background copy. The -cascade option has to be omitted because the
Global Copy is now not a cascaded relation.
dscli> mkpprc -dev IBM.2107-75ABTV2 -remotedev IBM.2107-75ABTV1 -type gcp -mode nocp
6400-6403:6200-6203
xecuting command:
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 6400:6200 successfully
created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 6401:6201 successfully
created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 6402:6202 successfully
created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 6403:6203 successfully
created.
dscli>
dscli> lspprc -dev IBM.2107-75ABTV2 -remotedev IBM.2107-75ABTV1 -l -fullid -fmt default
6400-6403
ID State Reason Type Out Of Sync
Tracks Tgt Read Src Cascade Tgt Cascade Date Suspended SourceLSS Timeout (secs)
Critical Mode First Pass Status
===========================================================================================
===========================================================================================
===========================
IBM.2107-75ABTV2/6400:IBM.2107-75ABTV1/6200 Copy Pending - Global Copy 0
Disabled Enabled invalid - IBM.2107-75ABTV2/64 unknown Disabled
True
IBM.2107-75ABTV2/6401:IBM.2107-75ABTV1/6201 Copy Pending - Global Copy 0
Disabled Enabled invalid - IBM.2107-75ABTV2/64 unknown Disabled
True
IBM.2107-75ABTV2/6402:IBM.2107-75ABTV1/6202 Copy Pending - Global Copy 0
Disabled Enabled invalid - IBM.2107-75ABTV2/64 unknown Disabled
True
IBM.2107-75ABTV2/6403:IBM.2107-75ABTV1/6203 Copy Pending - Global Copy 0
Disabled Enabled invalid - IBM.2107-75ABTV2/64 unknown Disabled
True
Example 27-12 shows the setup of the session and the Global Mirror and how to check if the
Global Mirror is running properly.
406 IBM System Storage DS8000 Series: Copy Services in Open Environments
28
In a planned failover the application has to be stopped before recovery at the intermediate
site. The host is then given access to the volumes at the intermediate site and the application
can be started using these volumes. A reason for a planned recovery at the intermediate site
could be because of maintenance activities at the local site that impact production (see 27.2,
“General consideration for storage failover” on page 396). A recovery at the intermediate site
minimizes the impact to the production normally running at the local site.
In a large data center many applications running on different servers may participate to the
Metro/Global Mirror. In this case it is most likely that failover operations will not be
accomplished against the complete environment but rather to specific applications. In this
case the failover operations have to be applied in a way that will not affect the copy relations
for the remaining applications. This situation is accounted for in all the scenarios presented in
this chapter.
Also, all the secnarios presented here have been practically tested and represent the best
practice for the situations they address. However, additional or alternate scenarios remain
possible depending on particular circumstances within your data center.
Note: If other scenarios are required, we strongly recommend that you extensively test
them before they are exercised in the production environment.
When the servers have a connectivity from the local site to the storage at the intermediate
site it is also possible to start the applications at the local site. This makes sense when the
maintenance operations at the local site only affect the storage system, but not the servers. In
this case the recovery process for the application becomes simpler.
A recovery at the intermediate site is in fact a simple failover feature of the Metro Mirror. The
Global Mirror remains untouched and continues to copy data to the remote site.
408 IBM System Storage DS8000 Series: Copy Services in Open Environments
Local Site Intermediate Site Remote Site
2 3
A B C
1
4 D
The steps to failover the production to the intermediate site are as follows:
1. Stop the production application at the local site.
2. Suspend the Metro Mirror.
3. Failover the Metro Mirror to the intermediate site.
4. Start the application at the intermediate site.
Example 28-1 illustrates how to bring the Metro Mirror pair into the suspended state. Before
the failoverpprc command is issued, the status of the volume pairs at the local and the
intermediate sites will be checked.
410 IBM System Storage DS8000 Series: Copy Services in Open Environments
Note: To failover to the intermediate site you have to specify the secondary volumes as the
source and the primary volumes as the target in the failoverpprc command.
Example 28-2. shows how to failover to the intermediate site using the failoverpprc
command.
Note that after a failover, the status of the Metro Mirror secondary volumes used in a
Metro/Global Mirror is different than the status of the failover in a conventional Metro Mirror. In
a conventional Metro Mirror the secondary volumes will be in the status Suspended with
reason Host Source. Because in a Metro/Global Mirror the Metro Mirror secondary volumes
are still cascaded to the remote volumes, the status of these volumes will not be Suspended
Host Source. In the current implementation the status of the secondary volumes is displayed
as Target Suspended with the reason Host Target.
The return from the intermediate site to the local site requires that the Global Copy must be
suspended. This is because the volumes cannot be the source for the remote and the local at
the same time. This enables two possibilities for the applications at the intermediate site that
have to be stopped:
1. When the data at the local site is too old because the production was kept for a relatively
long time at the intermediate site, it is safer to rely on the data in the Global Mirror. The
Global Mirror was always active, and consistent data is most current at the remote site. In
case of a problem during the failback, the data can be recovered from the Global Mirror.
For the failback process the applications should be stopped before the failback is issued.
The application can be started up at the local site, when the volume status of the Metro
Mirror is Full Duplex.
2. If the application downtime must be as short as possible, the applications may be stopped
after the failback to the local site, when all volumes are in Full Duplex. But this requires to
terminate the Global Copy. During the time when the intermediate volumes are
synchronized with the local volumes, the applications have no valid copy.
In Figure 28-2 the steps to move back to local is illustrated, whereby the application is
stopped first, to ensure that there is always a valid copy of the data available.
A B C
5 4
9 2 1 D
10
Disaster
Production
Recovery
Host
Test
Host
412 IBM System Storage DS8000 Series: Copy Services in Open Environments
6. Fail over to the local site.
7. Fail back Metro Mirror from the local site to the intermediate site.
8. Resume Global Copy.
9. Start I/O at the local site.
10.Start Global Mirror or add volumes to the session.
28.3.2 Step 2: Terminate Global Mirror or remove volumes from the session
Typically Metro/Global Mirror is deployed in an environment with more than one host
connected to the Metro/Global Mirror storage. Terminate Global Mirror session in case all
hosts are failed over to the intermediate site. If all hosts are not failed over to the intermediate
site, during the failback to the local site, the Global Mirror should be kept up and running to
ensure that the data for those hosts which remain at the local site is still copied in a consistent
way.
Prior to the failback the volumes belonging to the host which will failback, have to be removed
from the session. When this is done the Global Copy for these volumes can be suspended.
See Example 28-3.
Example 28-3 Remove the volumes from the session and pause the Global Copy
dscli> chsession -dev IBM.2107-75ABTV2 -lss 72 -action remove -volume 7200-7203 1
Date/Time: June 20, 2006 2:25:12 PM CEST IBM DSCLI Version: 5.1.600.196 DS:
IBM.2107-75ABTV2 CMUC00147I chsession: Session 1 successfully modified.
Example 28-5 Fail back to the local site and check the status
dscli> failbackpprc -dev IBM.2107-75ABTV2 -remotedev IBM.2107-75ABTV1 -type mmir
7200-7203:7000-7003
Date/Time: June 20, 2006 2:29:43 PM CEST IBM DSCLI Version: 5.1.600.196 DS:
IBM.2107-75ABTV2
CMUC00197I failbackpprc: Remote Mirror and Copy pair 7200:7000 successfully failed back.
CMUC00197I failbackpprc: Remote Mirror and Copy pair 7201:7001 successfully failed back.
CMUC00197I failbackpprc: Remote Mirror and Copy pair 7202:7002 successfully failed back.
CMUC00197I failbackpprc: Remote Mirror and Copy pair 7203:7003 successfully failed back.
dscli> lspprc -dev IBM.2107-75ABTV2 -remotedev IBM.2107-75ABTV1 -fullid -fmt default
7200-7203
Date/Time: June 20, 2006 2:30:53 PM CEST IBM DSCLI Version: 5.1.600.196 DS:
IBM.2107-75ABTV2
ID State Reason Type SourceLSS
Timeout (secs) Critical Mode First Pass Status
===========================================================================================
===================================================
IBM.2107-75ABTV2/7200:IBM.2107-75ABTV1/7000 Full Duplex - Metro Mirror
IBM.2107-75ABTV2/72 300 Disabled Invalid
IBM.2107-75ABTV2/7201:IBM.2107-75ABTV1/7001 Full Duplex - Metro Mirror
IBM.2107-75ABTV2/72 300 Disabled Invalid
IBM.2107-75ABTV2/7202:IBM.2107-75ABTV1/7002 Full Duplex - Metro Mirror
IBM.2107-75ABTV2/72 300 Disabled Invalid
IBM.2107-75ABTV2/7203:IBM.2107-75ABTV1/7003 Full Duplex - Metro Mirror
IBM.2107-75ABTV2/72 300 Disabled Invalid
28.3.5 Steps 5 and 6: Suspend Metro Mirror and fail over to the local site
When all the volumes at the local site are synchronized with the intermediate site, the Metro
Mirror must be failed over to the local site before the applications can be started.
In Example 28-6 prior to the failover the Metro Mirror is suspended by a pausepprc command
according to the recommendation in 27.5, “Suspend volumes before failover” on page 399. At
this point the I/O to the volumes has been stopped and the direction of the mirror will be
changed with the next failback.
Note: To failover to the intermediate site, you have to specify the secondary volumes as
the source and the primary volumes as targets in the failoverpprc command.
414 IBM System Storage DS8000 Series: Copy Services in Open Environments
CMUC00157I pausepprc: Remote Mirror and Copy volume pair 7203:7003 relationship
successfully paused.
dscli> failoverpprc -dev IBM.2107-75ABTV1 -remotedev IBM.2107-75ABTV2 -type mmir
7000-7003:7200-7203
Date/Time: June 20, 2006 2:36:04 PM CEST IBM DSCLI Version: 5.1.600.196 DS:
IBM.2107-75ABTV1
CMUC00196I failoverpprc: Remote Mirror and Copy pair 7000:7200 successfully reversed.
CMUC00196I failoverpprc: Remote Mirror and Copy pair 7001:7201 successfully reversed.
CMUC00196I failoverpprc: Remote Mirror and Copy pair 7002:7202 successfully reversed.
CMUC00196I failoverpprc: Remote Mirror and Copy pair 7003:7203 successfully reversed.
28.3.6 Step 7: Fail back Metro Mirror from the local site to the intermediate site
The failback from the local to the intermediate site triggers the Metro Mirror to copy the tracks
that are out-of-sync immediately after I/O to the primary volumes has been started.
Example 28-7 shows the failback to the intermediate site. This command is executed at the
local site.
Example 28-7 Fail back from the local to the intermediate site
dscli> failbackpprc -dev IBM.2107-75ABTV1 -remotedev IBM.2107-75ABTV2 -type mmir
7000-7003:7200-7203
Date/Time: June 20, 2006 2:37:09 PM CEST IBM DSCLI Version: 5.1.600.196 DS:
IBM.2107-75ABTV1
CMUC00197I failbackpprc: Remote Mirror and Copy pair 7000:7200 successfully failed back.
CMUC00197I failbackpprc: Remote Mirror and Copy pair 7001:7201 successfully failed back.
CMUC00197I failbackpprc: Remote Mirror and Copy pair 7002:7202 successfully failed back.
CMUC00197I failbackpprc: Remote Mirror and Copy pair 7003:7203 successfully failed back.
28.3.9 Step 10: Start Global Mirror or add volumes to the session
In order to form Consistency Groups, start the Global Mirror session if it was terminated (see
28.3.2, “Step 2: Terminate Global Mirror or remove volumes from the session” on page 413)
or add the volumes to the session again if the Global Mirror is still up and running (as would
be the case when other applications that are not the subject of the failover/failback scenarios
are present and and need to have the Global Mirror kept in operation).
The volumes are added to the session with the chsession command using the option -action
add, as shown in Example 28-9. When done, the volume status will be Join Pending as long
as the first pass copy of the Global Copy has not been finished. Use the lssession command
to query the staus of the volumes in an LSS.
Example 28-9 Add volumes to the session and check the session
dscli> chsession -dev IBM.2107-75ABTV2 -lss 72 -action add -volume 7200-7203 1
Date/Time: June 20, 2006 2:41:38 PM CEST IBM DSCLI Version: 5.1.600.196 DS:
IBM.2107-75ABTV2
CMUC00147I chsession: Session 1 successfully modified.
dscli> lssession -dev IBM.2107-75ABTV2 -fmt default 72
Date/Time: June 20, 2006 2:42:18 PM CEST IBM DSCLI Version: 5.1.600.196 DS:
IBM.2107-75ABTV2
LSS ID Session Status Volume VolumeStatus PrimaryStatus SecondaryStatus
FirstPassComplete AllowCascading
===========================================================================================
==================================
72 01 CG In Progress 7200 Active Primary Copy Pending Secondary Full
Duplex True Enable
72 01 CG In Progress 7201 Active Primary Copy Pending Secondary Full
Duplex True Enable
72 01 CG In Progress 7202 Active Primary Copy Pending Secondary Full
Duplex True Enable
72 01 CG In Progress 7203 Active Primary Copy Pending Secondary Full
Duplex True Enable
416 IBM System Storage DS8000 Series: Copy Services in Open Environments
Figure 28-3 illustrates the steps it takes to recover to the remote site. This scenario includes
the setup of a Global Mirror from the remote site to the intermediate site in order to be
prepared for keeping the production at the remote site for an extended period of time.
Therefore the data will be replicated to the intermediate site.
4 5 6
A B C
3
1 2 7 D
Secondary
Primary production
production host
host
When the application will remain for an extended period of time at the remote site and the
intermediate is still available, it is possible to create an additional Global Mirror from the
remote site to the intermediate site in order to protect the data against a possible disaster at
the remote site. See 27.8, “Set up an additional Global Mirror from remote site” on page 403,
for the details.
The recovery of the production to the remote site comprises the following steps:
1. Stop I/O at the local site.
2. Terminate Global Mirror or remove volumes from the session.
3. Terminate Global Copy.
4. Suspend Metro Mirror.
5. Fail over Metro Mirror to the intermediate site.
6. Establish Global Copy from the remote site to the intermediate site.
7. Start I/O at the remote site.
Attention: Between the current step (step 3) and up through step 6, make sure that all I/O
to the volumes are suspended. If not, data may be corrupted due to the no copy option,
which would then require a full copy.
Important: Before the Global Copy is removed, check that all tracks were copied to the
secondary site. Use the command lspprc -l for obtaining the out-of-sync tracks.
418 IBM System Storage DS8000 Series: Copy Services in Open Environments
CMUC00155I rmpprc: Remote Mirror and Copy volume pair 7202:7402 relationship successfully
withdrawn.
CMUC00155I rmpprc: Remote Mirror and Copy volume pair 7203:7403 relationship successfully
withdrawn.
Important: Before the Metro Mirror is suspended, check that all tracks were copied to the
intermediate site. Use the lspprc -l command for obtaining the out-of-sync tracks.
Note: If the links between the local and the intermediate site have failed, suspending the
Metro Mirror is not applicable, because the pair status has been turned into suspended.
Example 28-1 on page 409 illustrates how to bring the Metro Mirror pair into the suspended
state. Before the failoverpprc command is issued, the status of the volume pairs at the local
and the intermediate site must be checked.
The Metro Mirror failover can also be seen as a preparation for the failback to the local site.
The reversed Metro Mirror will act as a cascade from the reversed Global Copy between the
remote and intermediate sites. For this reason the Metro Mirror secondary volumes failover is
issued with the -cascade option. Also, to avoid extended response times at the host side, a
cascaded copy relation should not be set up as a synchronous relation. Thus, the
failoverpprc command for the Metro Mirror will be executed with the option -type gcp.
Note: To fail over to the intermediate site you must specify the secondary volumes as the
source and the primary volumes as targets in the failoverpprc command.
If at the intermediate site additional volumes can be provided for a new FlashCopy relation to
the intermediate volumes, it is possible to set up an additional Global Mirror. Assuming that
the production may remain for a longer period of time at the remote site, this approach would
provide consistent data at the intermediate site and protect the production against a possible
disaster at the remote site. See 27.8, “Set up an additional Global Mirror from remote site” on
page 403, for details about how to set it up.
Example 28-14 shows how to establish the Global Copy. The -cascade option has to be
omitted because the Global Copy is now not a cascaded relation anymore. Since the volumes
at the remote and intermediate sites are identical, use a -mode nocp option in order to avoid a
FULL COPY.
420 IBM System Storage DS8000 Series: Copy Services in Open Environments
28.5 Return from remote site
The scenario described in this section includes the return from the remote site after a planned
recovery, which was discussed in 28.4, “Recovery at remote site” on page 416. This scenario
also applies after a recovery at the remote site has taken place due to a failure situation at the
local site. If the recovery at the remote site has been accomplished because of a failure
situation at the local site, the return to the local site can only be accomplished when all
required resources are available again.
As part of the recovery at the remote site scenario a Global Copy was established from the
remote to the intermediate site to ensure that the data was copied to the intermediate site
while production was running at the remote site. If this step had been omitted it must be
applied now. Ensure that all data has been drained to the intermediate site and then to the
local site before the applications are started again at the local site.
If an additional Global Mirror has been set up as described in 27.8, “Set up an additional
Global Mirror from remote site” on page 403, then it has to be removed before you proceed to
perform the steps to return to the local site.
6 7
A B C
5 4 2 3
9 D
8
1
Primary Secondary
prodution production
Host host
28.5.2 Step 2: Fail back Metro Mirror from the intermediate site to the local site
The running application’s data at the remote site has been replicated because a Global Copy
was established during the failover procedure (see 28.4.6, “Step 6: Establish Global Copy
from remote to intermediate site” on page 420). When the local site becomes available again,
the Metro Mirror can be failed back.
The Metro Mirror failover described under 28.4.5, “Step 5: Failover Metro Mirror to
intermediate site” on page 419, must be issued with the option -cascade since with the fail
back to the local site, the Metro Mirror is now cascaded again to the Global Copy. This also
implies that a cascaded copy relation to the local site must not be a synchronous relation.
This is realized with the option -type gcp, which turns the Metro Mirror into Global Copy mode.
If the failover was not executed with these two options during the failover to the remote site
scenario, these options can be supplied right now with the failbackpprc command.
In Example 28-15 the failbackpprc command is issued including the options -cascade and
-type gcp. It is executed at the intermediate site. Before taking any further action, make sure
that all data was replicated to the local site. Check this with the lspprc command.
422 IBM System Storage DS8000 Series: Copy Services in Open Environments
(see 28.5.7, “Step 7: Create Global Copy from intermediate to remote site” on page 424) the
Global Copy is recreated from the intermediate site to the remote site in nocopy mode.
Attention: Make sure that between this step and step 6 no I/O will happen to the volumes.
Otherwise data may be corrupted due to the no copy option, and thus a full copy is
required.
Example 28-16 shows the DS CLI command to remove the Global Copy.
Important: Before the Global Copy is removed, check whether all tracks are copied to the
secondary site. Use the TSO CQUERY command to verify whether the Percent of copy
complete is 100%.
Example 28-16 Remove the Global Copy from remote to intermediate site
dscli> rmpprc -quiet -dev IBM.2107-75BYGT1 -remotedev IBM.2107-75ABTV2 7400-7403:7200-7203
Date/Time: June 21, 2006 12:44:33 PM CEST IBM DSCLI Version: 5.1.600.196 DS:
IBM.2107-75BYGT1
CMUC00155I rmpprc: Remote Mirror and Copy volume pair 7400:7200 relationship successfully
withdrawn.
CMUC00155I rmpprc: Remote Mirror and Copy volume pair 7401:7201 relationship successfully
withdrawn.
CMUC00155I rmpprc: Remote Mirror and Copy volume pair 7402:7202 relationship successfully
withdrawn.
CMUC00155I rmpprc: Remote Mirror and Copy volume pair 7403:7203 relationship successfully
withdrawn.
Example 28-17 shows how to fail over to the local site. The command is executed at the
intermediate site.
Note: To fail over to the intermediate site you must specify the secondary volumes as the
source and the primary volumes as targets in the failoverpprc command.
28.5.6 Step 6: Fail back Metro Mirror from the local site to the intermediate site
The failback enables tracks to be copied from the local to the intermediate site. The Metro
Mirror is now no longer in a cascaded relation, so the copy type is now mmir again. Because
the applications have not yet been started, the content of the volume at the local and the
intermediate site are the same, and the status of the Metro Mirror should be Full Duplex.
Example 28-18 shows the failback to the intermediate site at the local site.
Example 28-18 Fail back the Metro Mirror from the local site to the intermediate site
dscli> failbackpprc -dev IBM.2107-75ABTV1 -remotedev IBM.2107-75ABTV2 -type mmir
7000-7003:7200-7203
Date/Time: June 21, 2006 12:49:33 PM CEST IBM DSCLI Version: 5.1.600.196 DS:
IBM.2107-75ABTV1
CMUC00197I failbackpprc: Remote Mirror and Copy pair 7000:7200 successfully failed back.
CMUC00197I failbackpprc: Remote Mirror and Copy pair 7001:7201 successfully failed back.
CMUC00197I failbackpprc: Remote Mirror and Copy pair 7002:7202 successfully failed back.
CMUC00197I failbackpprc: Remote Mirror and Copy pair 7003:7203 successfully failed back.
dscli> lspprc -dev IBM.2107-75ABTV1 -remotedev IBM.2107-75ABTV2 -fullid -fmt default
7000-7003
Date/Time: June 21, 2006 12:50:59 PM CEST IBM DSCLI Version: 5.1.600.196 DS:
IBM.2107-75ABTV1
ID State Reason Type SourceLSS
Timeout (secs) Critical Mode First Pass Status
===========================================================================================
===================================================
IBM.2107-75ABTV1/7000:IBM.2107-75ABTV2/7200 Full Duplex - Metro Mirror
IBM.2107-75ABTV1/70 300 Disabled Invalid
IBM.2107-75ABTV1/7001:IBM.2107-75ABTV2/7201 Full Duplex - Metro Mirror
IBM.2107-75ABTV1/70 300 Disabled Invalid
IBM.2107-75ABTV1/7002:IBM.2107-75ABTV2/7202 Full Duplex - Metro Mirror
IBM.2107-75ABTV1/70 300 Disabled Invalid
IBM.2107-75ABTV1/7003:IBM.2107-75ABTV2/7203 Full Duplex - Metro Mirror
IBM.2107-75ABTV1/70 300 Disabled Invalid
Important: Be sure that there is no active I/O to the volumes between step 3 and step 7.
Otherwise data may be corrupted due to the nocopy option and would then require a full
copy.
424 IBM System Storage DS8000 Series: Copy Services in Open Environments
Example 28-19 shows the DS CLI command to create the Global Copy.
Example 28-19 Create Global Copy from the intermediate site to the remote site
dscli> mkpprc -dev IBM.2107-75ABTV2 -remotedev IBM.2107-75BYGT1 -type gcp -mode nocp
-cascade 7200-7203:7400-7403
Date/Time: June 21, 2006 12:53:07 PM CEST IBM DSCLI Version: 5.1.600.196 DS:
IBM.2107-75ABTV2
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 7200:7400 successfully
created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 7201:7401 successfully
created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 7202:7402 successfully
created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 7203:7403 successfully
created.
Example 28-20 shows how to add the volumes into the session and how to monitor the
session. Because the applications have not been started, the volumes should go immediately
into the active state.
Also, all of the secnarios presented here have been practically tested and represent the best
practice for the situations they address. However, additional or alternate scenarios remain
possible depending on particular circumstances within your data center.
Note: We strongly recommend that you test all procedures extensively before deployment
to the production environment.
Figure 29-1 illustrates the steps to perform a failover to the intermediate site, while the
production remains at the local site.
A B C
1
2 E D
Production 5
Host
Disaster
Recovery
Test
Host
The steps for this scenario of the failover to the intermediate site are as follows:
1. Prepare the failover by issuing a freeze and unfreeze of the Metro Mirror.
Note: During the freeze/unfreeze interval no data will be copied to the remote site.
428 IBM System Storage DS8000 Series: Copy Services in Open Environments
2. Establish FlashCopy to the additional volumes at the intermediate site.
3. Set up PPRC paths from the local site to the intermediate site.
4. Resume Metro Mirror.
5. Start I/O at the disaster recovery host.
Note: During the queue full state the I/O from the application is blocked and waits until the
volumes are writable again. This implies an interruption of the application I/O for the time it
takes to terminate the PPRC links. The application does not have to be stopped.
Example 29-1 shows the usage of the freezepprc and the unfreezepprc for the Metro Mirror.
Theses commands are issued at the local site. A subsequent lspprc command at the local
storage subsystem shows that the primary volumes went to a Suspended state as a result of
Freeze Metro Mirror. An lspprc command issued at the intermediate site shows the status of
the secondary volumes, which are still in Target Full Duplex mode.
To quickly reenable I/O to the primary volumes, the unfreezepprc command is issued
immediately after the freezepprc. Example 29-2 illustrates the unfreezepprc command.
Example 29-2 Unfreeze the primary volumes and recreate PPRC paths
dscli> unfreezepprc -dev IBM.2107-75ABTV1 -remotedev IBM.2107-75ABTV2 70:72
Date/Time: June 22, 2006 11:43:00 AM CEST IBM DSCLI Version: 5.1.600.196 DS:
IBM.2107-75ABTV1
CMUC00198I unfreezepprc: Remote Mirror and Copy pair 70:72 successfully thawed.
Example 29-3 shows the mkflash command, which is executed at the intermediate site.
430 IBM System Storage DS8000 Series: Copy Services in Open Environments
29.2.3 Step 3: Set up PPRC paths from the local site to the intermediate site
After the FlashCopy has been initiated, we can set up the PPRC paths again. More details
about the PPRC paths set up can be found in 26.3.2, “Step 1: Set up all Metro Mirror and
Global Mirror paths” on page 386. When the paths are recreated the Metro Mirror will not
copy data because the primary volumes are still in Suspend mode.
Example 29-4 shows the mkpprcpath and lspprcpath commands, which are executed on the
local site.
Example 29-4 Set up PPRC paths from the local site to the intermediate site and check the paths
When you have finished the disaster recovery test, you can unmount the file systems and
vary off the volume groups on the host and remove your FlashCopy target volumes at the
intermediate DS8000, so you are in the same state you were at the beginning of the disaster
recovery test.
432 IBM System Storage DS8000 Series: Copy Services in Open Environments
Figure 29-2 illustrates the steps to perform a failover to the remote site, while the production
remains at the local site.
3 4
A B C
D
2
1
E
5
Disaster
Recovery
Prodution Test
Host Host
Note: During the freeze/unfreeze interval no data will be copied to the remote site.
For this scenario it is assumed that an additional host is available to start the application from
the additional volumes that have been copied with FlashCopy at the remote site.
The steps are explained in detail in 29.2, “Disaster recovery test at the intermediate site” on
page 428. The only difference is that steps 2 and 5 will be executed on the remote site.
Depending on the kind of incident, different actions are required. It is therefore essential to
analyze the current situation very carefully in order to make the right decisions for the storage
takeover. When the situation is understood, actions should be managed in the way that one
of the planned scenarios described in Chapter 28, “Planned recovery scenarios” on
page 407, can be used to bring the system into a defined failover status.
Some situations are easier to manage when a disaster occurs. For example, when the local
site is completely destroyed, it is obvious that the storage and the applications have to be
failed over to the intermediate or the remote site. Other situations may be more complex, for
example, when some components of the local site are still in operation and other components
are damaged or inaccessible. In this case a careful analysis of the situation has to be done
before the failover processes can be accomplished.
In the following sections several types of incidents and disaster situations are discussed,
though many variations are possible.
A server outage is, in general, addressed through a high availability solution like HACMP™
for AIX or other cluster solutions implemented at the local site. When the intermediate site is
not too far away from the local site and the bandwidth is high enough, a cluster across the
local and the intermediate site can be implemented as described in 26.2, “Configuration
examples” on page 383.
436 IBM System Storage DS8000 Series: Copy Services in Open Environments
When there is no cluster implementation and a server outage happens, the situation can be
categorized as a disaster situation where the application and the storage have to be taken
over. This is the situation discussed in this section and illustrated in Figure 30-1.
A B C
Failover
Primary Secondary
production production
Host host
In this case the storage focal point is failed over to the remote site, since the second server
will typically reside at the remote site. To achieve the failover to the remote site the scenario
described in 28.4, “Recovery at remote site” on page 416, can be used.
Using the Incremental Resync functionality it is possible to set up a Global Mirror to the
remote site, which bypasses the intermediate site, without any interruption of the data
replication. See Chapter 31, “MGM Incremental Resync” on page 443, for more information.
A B C
Primary Secondary
production production
Host host
When all links have failed in a Metro Mirror, the pairs will go into the suspend mode after a
time-out of 30 seconds (which is the default).
To fail over to the intermediate site, see 28.2.3, “Step 3: Failover the intermediate site” on
page 410, and proceed with this step.
When a failover to the remote site has to be done, because the secondary production server
is located at the remote site, the processing should start with 28.4.2, “Step 2: Terminate
Global Mirror or remove volumes from session” on page 418, and the following steps.
If the Metro Mirror links are back before a failover has been started, the Metro Mirror pairs
have to be resumed manually with the resumepprc command.
Redundancy provides an overall resilient environment for each data center, and is enough to
resist most of the incidents against single components like the servers, storage, network, or
parts of the infrastructure. But serious incidents, and usually beyond your control, such as a
438 IBM System Storage DS8000 Series: Copy Services in Open Environments
massive lightning stroke, may cause an outage of more than one component of the
infrastructure. Other parts of the data center may still be faultless.
Figure 30-3 illustrates an example of a partial disaster in a data center caused by multiple
outages of the electrical power distribution.
Secondary
power
distribution
360V
4kV
SD1
Primary
power
distribution
Server 1 Storage 1
SD2
20kV
PD1
Storage 2
SD3 Server 2
Storage 3
20kV SD4
PD1
SD5
SD6
In this example, three secondary power distributors are out of order. The result is that all
components that are connected to SD1 or SD4 only or connected to both SD1 and SD4 have
no electrical power. All other components continue to operate.
Assume that Server1 is configured into several Logical Partitions (LPARs) and that in each
LPAR a different application or a part of an application, for example, a database engine, is
running. The storage for server 1 may be supplied by the storage subsystem Storage1,
Storage2, and Storage3.
In order to make the correct decisions for recovery, a detailed and careful analysis of this
situation is essential. Remember that not only the servers and the storage subsystems are
the subject of the recovery, but that the application must be made available at the recovery
site.
With Metro/Global Mirror you can bring applications back very quickly, and in a flexible way,
while keeping data consistent.
Based on the conclusions made above, the following decision table may help you to find the
correct scenario for the Metro/Global Mirror operations.
440 IBM System Storage DS8000 Series: Copy Services in Open Environments
Server Storage Location of Apply Apply Applicable
status status secondary storage server scenario
server failover failover starting at
A partial disaster, as can occur in large data centers, is the most challenging kind of disaster
because of the many different situations that can result. The complexity is apparent even
when only 20% of the applications are affected. This percentage is too low to fail over a
complete data center. But when a data center hosts 100 applications, the client has to
manage 20 different critical situations.
For a Metro/Global Mirror implementation, disaster recovery is only indicated when there is an
outage of the data center at the local site. In this case an unplanned failover scenario like that
described in 30.3.1, “Metro Mirror link failure” on page 437, can be applied.
When the intermediate or the remote site is down, the production at the local site can still
continue. Only the disaster recovery capabilities are degraded. When the remote site is down,
Metro Mirror to the intermediate site is still available.
When the intermediate site is down and there is a connection with sufficient bandwidth
available between the local and the remote site, a Global Mirror to the remote site can be set
up. In this case a Global Copy between the Metro Mirror primary volumes and the Global
Copy secondary volumes can be set up. This requires that at the local and the remote site the
current pair relation is removed using the rmpprc command with the options -at src and
-unconditional. The links between the storage subsystems at the local and the remote site
have to be established and the paths have to be created according to 26.3.2, “Step 1: Set up
all Metro Mirror and Global Mirror paths” on page 386. The steps required to set up the Global
Mirror are described in 26.3.3, “Step 2: Set up Global Copy NOCOPY from intermediate to
remote sites” on page 388, and in 26.3.6, “Step 5: Create Global Mirror session and add
volumes to session” on page 389, and the following sections.
442 IBM System Storage DS8000 Series: Copy Services in Open Environments
31
The chapter starts with an overview and functional description of Incremental Resync and
explains the new DSCLI options that support it. This is followed by a section explaining how
to set up Incremental Resync for Metro/Global Mirror.
Finally, this chapter includes scenarios with detailed operations presented in a step-by-step
approach and covering the following situations:
Failure at the local site
Failure at the intermediate site
Returning to normal operations
Global Mirror
2109-M14
2109-M14
2109-M14
P590
P590
P550 P550
To enable the Incremental Resync capability a specific option must be specified when initially
establishing the Metro Mirror relationship between the local and intermediate sites. If the
Metro Mirror is already established and in full duplex, the Incremental Resync is enabled by
issuing the mkpprc command against the existing relation with the specific option to enable
Incremental Resync.
444 IBM System Storage DS8000 Series: Copy Services in Open Environments
When you specify Incremental Resync, in addition to the existing Change Recording bitmap
of the Global Mirror at the intermediate site, another Change Recording bitmap is created at
the Metro Mirror primary volumes. This bitmap keeps track of any incoming I/O while a
consistency group is copied from the intermediate site until it has arrived to the FlashCopy
target volumes at the remote site. The Global Mirror at the intermediate site is queried
periodically for the current status of the tracks between the intermediate and the remote site.
The change recording will restart from a cleared change recording bitmap, when the next
consistency group is formed at the intermediate site. Figure 31-2 shows an overview of the
different phases of the Incremental Resync.
C CO
R RO
Incremental Resync S
bitmap Global Mirror
bitmaps
querying
A B C
If there is a failure of the intermediate site, the Metro Mirror primary volumes will go into
suspend mode. With Incremental Resync enabled, the tracks that could not be copied to the
intermediate site are recorded in the change recording bitmap at the primary site.
To bypass the failed intermediate site, a new Global Copy relationship should now be
established from the local volumes to the FlashCopy source volumes at the remote site. At
this point in time, however, these volumes are still Global Copy target volumes of the relation
from the intermediate site. Therfore, failover of the existing Global Copy at the remote site
must take place before the new Global Copy between the local and remote sites can be
established.
A B C
Primary
production
host
When the intermediate site is available again the Metro/Global Mirror must be re-configured
back to normal operation.
First of all, the intermediate site has to be checked for its current status of the relations. The
former Global Mirror from the intermediate site to the remote site must be removed. To copy
data missing at the intermediate site since the failure, a failback from the remote site to the
intermediate site is accomplished.
When the incremental copy is completed, the procedure to return to normal operations from
local to remote via the intermediate site can be initiated.
First, the Global Mirror and the Global Copy relationships to the remote site have to be
removed. Next, the Global Copy between the intermediate and the remote site has to be
re-established in its original direction. Then the Metro Mirror can be recreated using the
Incremental Resync option. The Metro Mirror will copy only the data changed since the
Global Copy from the local to the remote site was removed. The final step is to recreate the
Global Mirror at the intermediate site, which enables consistency groups to be formed at the
intermediate and drained to the remote site.
446 IBM System Storage DS8000 Series: Copy Services in Open Environments
consistency groups are formed in Global Mirror the bitmaps will
change to reflect only the changed data.)
enablenoinit This enables the Incremental Resync functionality and creates change
recording bitmaps. The bitmaps will not be initialized. That is, they will
all be zeroes. This option should only be used in certain cases as part
of the return scenario for Incremental Resync.
disable This stops the Incremental Resync mechanism.
recover This brings in relations of local to remote volumes as new pairs. The
devices specified in the command are checked to ensure that they
currently have a target device in common. In other words, it verifies
that the devices are part of an existing Metro/Global Mirror
relationship.
override This is the same as above, but the volumes may not be part of a
Metro/Global Mirror. As opposed to a recover, in this case the
checking is not done
The steps must be executed in the same order. However, 26.3.4, “Step 3: Set up Metro Mirror
between local and intermediate sites” on page 388, is modified as follows to implement
Metro/Global Mirror with Incremental Resync.
Specify the -incrementalresync enable option in conjunction with -type mmir options when
executing the mkpprc command at the local site to create the Metro Mirror relationship. This
enables the Incremental Resync function and creates change recording bitmaps for the Metro
Mirror primary volumes running at the local site.
When the Metro Mirror is already established, but Incremental Resync is not yet enabled, the
option -mode nocp is required to enable Incremental Resync.
Example 31-1 shows how to set up incremental re-synchronization for Metro/Global Mirror.
The remaining steps in 26.3, “Initial setup of Metro/Global Mirror” on page 385, are followed
to complete the setup for Metro/Global Mirror with incremental resync.
Figure 31-4 illustrates the steps to transition from a current Global Mirror environment to a
Metro/Global Mirror with Incremental Resync environment.
Primary
production Local Site
host
A
3 Remote Site
4
1
7
2
Intermediate Site C
D
6 5
B 1
Figure 31-4 Moving from Global Mirror to Incremental Resync Metro/Global Mirror
448 IBM System Storage DS8000 Series: Copy Services in Open Environments
The steps for setting up and initializing Metro/Global Mirror with Incremental Resync if Global
Mirror is already running are as follows. (The assumption is that there is an existing Global
Mirror between the A and C volumes, and an intermediate site (B volumes) is added to the
configuration.)
1. Set up all PPRC paths.
2. Start Global Copy from the remote to the new intermediate site.
3. Start incremental re-synchronization from the local site.
4. Terminate Global Mirror and suspend Global Copy at the local site.
5. Terminate locate to remote Global Copy local to remote at remote site.
6. Reverse Global Copy to run from intermediate to remote site.
7. Set up Metro Mirror from local to intermediate site.
8. Set up Global Mirror at intermediate site.
Once the PPRC ports are identified, PPRC paths can then be created from the local to the
intermediate site and another from the intermediate to the remote site. It is also good practice
to create the PPRC paths in the opposite directions, from the intermediate to the local site
and from the remote to the intermediate site. There should be a total of four new PPRC paths
created. Prerequisites for creating a successful PPRC path and reasons for creating PPRC in
both directions are explained in 26.3.2, “Step 1: Set up all Metro Mirror and Global Mirror
paths” on page 386.
To create the new PPRC paths, the command mkpprcpath is executed from each of the sites.
Step 2: Start Global Copy from the remote to the intermediate site
To migrate the data from the remote site to the intermediate site, Global Copy is started. This
will take the data that is currently running from the local to remote and copy it (or rather
cascade it) to the intermediate site.
The mkpprc command with the option -type gcp is executed from the remote site to the
intermediate site. Because there is already a Global Copy relationship from the local to the
remote site, the -cascade option must also be used with the mkpprc command. This command
will create a Global Copy relationship from the remote site volumes that are currently Global
Copy secondaries to the intermediate site volumes.
Example 31-2 shows the DSCLI command to create the Global Copy. The subsequent lspprc
command shows the cascading to from local to remote to intermediate.
Example 31-2 Start Global Copy from the remote to the intermediate site
dscli> mkpprc -dev IBM.2107-75DNXC1 -remotedev IBM.2107-7520781 -type gcp -mode full -cascade 6000-6003:6600-6603
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 6000:6600 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 6001:6601 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 6002:6602 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 6003:6603 successfully created.
dscli>
dscli> lspprc -dev IBM.2107-75DNXC1 -fullid 6000-6003
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First
Pass
Status====================================================================================================================================
=================
IBM.2107-7503461/6200:IBM.2107-75DNXC1/6000 Target Copy Pending - Global Copy IBM.2107-7503461/62 unknown Disabled
Invalid
IBM.2107-7503461/6201:IBM.2107-75DNXC1/6001 Target Copy Pending - Global Copy IBM.2107-7503461/62 unknown Disabled
Invalid
The mkpprc command with the option -incrementalresync enablenoinit is executed from the
local site using the current Global Mirror relationship from the local to the remote site as
shown in Example 31-3.
The pausepprc command is used to suspend the Global Copy relationship from the local site
to the remote site. By suspending the Global Copy relationship, the local and remote site
volumes will both be in a suspended state and the change recording bitmaps will remain
active at the local site.
By suspending Global Copy from the local to the remote site, data at the intermediate site will
no longer be in sync with data at the local site. Once Global Copy is suspended, data is no
longer moving to the intermediate site via the remote site. Therefore the data at the
intermediate site begins to age as production continues running at the local site. The change
recording bitmaps at the local site are keeping track of all updates from production.
450 IBM System Storage DS8000 Series: Copy Services in Open Environments
In Example 31-4 the Global Mirror and the sessions are removed at the local site. Finally the
Global Copy from local to remote is suspended.
Example 31-4 Terminate Global Mirror and suspend Global Copy at local
dscli> rmgmir -dev IBM.2107-7503461 -quiet -lss 62 -session 2
CMUC00165I rmgmir: Global Mirror for session 2 successfully stopped.
dscli> chsession -dev IBM.2107-7503461 -lss 62 -action remove -volume 6200-6203 2
CMUC00147I chsession: Session 10 successfully modified.
dscli> rmsession -dev IBM.2107-7503461 -quiet -lss 62 2
CMUC00146I rmsession: Session 2 closed successfully.
dscli>
dscli> pausepprc -dev IBM.2107-7503461 -remotedev IBM.2107-75DNXC1 6200-6203:6000-6003
CMUC00157I pausepprc: Remote Mirror and Copy volume pair 6200:6000 relationship successfully paused.
CMUC00157I pausepprc: Remote Mirror and Copy volume pair 6201:6001 relationship successfully paused.
CMUC00157I pausepprc: Remote Mirror and Copy volume pair 6202:6002 relationship successfully paused.
CMUC00157I pausepprc: Remote Mirror and Copy volume pair 6203:6003 relationship successfully paused.
dscli>
dscli>
This step is necessary to allow the failback of the intermediate to the remote site when
reversing Global Copy in the next step“Step 6: Reverse Global Copy to run from intermediate
to remote site” on page 452.
Note: All remaining out-of-sync tracks are transferred at this point from the remote to the
intermediate site and must be drained before proceeding to the next step. To query
out-of-sync tracks, issue the command lspprc -l at the remote site. Once the out-of-sync
tracks are drained, the remote site and the intermediate site will be in sync.
In Example 31-5 the DSCLI command to terminate the Global Copy at the remote site is
shown. The lspprc command at the local site shows that the relations at this site are
untouched, while at the remote site only the Global Copy from remote to intermediate exists.
To fail over to the intermediate, the failoverpprc command with the option -type gcp is
executed at the intermediate site. Because the Global Copy that is currently running from the
remote to the intermediate is cascaded, the -cascade option is also specified with the
failoverpprc command. Once the failoverpprc command is executed with the -type gcp
and -cascade options, the remote site volumes status will become primary suspended.
Next is to fail back Global Copy at the intermediate site by using the failbackpprc command
with the option -type gcp at the intermediate site. Once again, since Global Copy is in a
cascaded relation, the -cascade option is also specified.
Failover and failback has successfully reversed the Global Copy relationship. Now Global
Copy can be started from the intermediate site. To start Global Copy, the mkpprc command
with the -type gcp and -cascade options is executed. Once again, the -cascade option is used
in preparation for the next step, where the Global Copy primary is also going to be a Metro
Mirror secondary.
Step 7: Set up Metro Mirror from the local to the intermediate site
Metro Mirror from the local to the intermediate site can now be created. The command mkpprc
with the -type mmir option is executed at the local site. In “Step 3: Start incremental
re-synchronization from the local site” on page 450 change recording bitmaps were created
to keep track of all updates from production. To recover and restore these updates, the
-incrementalresync recover option is specified with the mkpprc command. The recover
parameter of the -incrementalresync will establish the Metro Mirror relationship after doing a
check at the intermediate site (the Metro Mirror secondary) for a relationship. The change
recording bitmaps created in “Step 3: Start incremental re-synchronization from the local site”
on page 450 are then merged with the out-of-sync bitmaps at the local site.
Note: To prepare for a disaster, the next step would be to start Metro Mirror with
Incremental Resync enabled. The command mkpprc with the options -type mmir and
-incrementalresync enable would be executed at the local site. This creates new change
recording bitmaps for all the Metro Mirror primary volumes.
Note: Metro Mirror will need to become full duplex and Global Copy will need to complete
first pass. To query the status of Global Mirror and Metro Mirror, use the lspprc -l
command at the local site to query Metro Mirror and at the intermediate site to query
Global Copy.
452 IBM System Storage DS8000 Series: Copy Services in Open Environments
In Example 31-6 the Metro Mirror created in the first step with the option -incrementalresync
recover, which initiates copying all tracks that are marked in the change recording bitmap at
the local site. Once all the tracks have drained to the intermediate site the incremental
re-synchronization is restarted.
Important: Verify that the out-of-sync tracks have completed for Metro Mirror before
starting Global Mirror. Otherwise the consistency groups will start to fail when Global Mirror
is started with the mkgmir command.
Example 31-7 shows the steps to setup the Global Mirror. The showgmir commands at the
end verify that the Global Mirror is forming consistency groups.
454 IBM System Storage DS8000 Series: Copy Services in Open Environments
Moving production back to the local site and going back to the original Metro/Global Mirror
configuration.
Local Site
Remote Site
A
Primary
production
1
host
2
C
Intermediate Site
Secondary
production D
host
The following steps must be performed to transition to a two- site environment after the local
site failure:
1. Fail over to the remote site.
2. Terminate Metro Mirror at the intermediate site.
3. Start applications at the intermediate site.
In Example 31-9 the DSCLI command to terminate the Metro Mirror at the intermediate site is
shown. With the following lspprc command issued at the intermediate site, you can see that
only the relation to the remote site remains.
456 IBM System Storage DS8000 Series: Copy Services in Open Environments
Step 3: Start applications to the intermediate site
Production can now be moved from the local site to the intermediate site and applications can
be started. The out-of-sync bitmaps will have continued to be in use. Global Mirror will
continue to operate from the intermediate site to the remote site until the local site is brought
back and made available again.
Figure 31-6 illustrates the steps to start up Metro/Global Mirror with Incremental Resync
when the local site is available to use.
Local Site
Remote Site
1
8 A 4
C
Intermediate Site 7
D
3
B
Secondary
production
host
The following steps are taken when re-introducing the local site and moving back to a
Metro/Global Mirror with Incremental Resync environment:
1. Fail back Global Copy from remote to local site.
2. Start incremental re-synchronization at the intermediate site.
3. Terminate Global Mirror and suspend Global Copy from the intermediate to the remote
site.
4. Suspend Global Copy from the remote site to the local site.
5. Terminate Global Copy at the remote site.
Note: Remote to local site out-of-sync tracks will need to be drained completely before
continuing to the next step. This will ensure that the local site has been fully
re-synchronized.
Tip: When the local site is back, in addition to cleaning up the remaining relations, it may
also be necessary to clear remaining SCSI reservations from the host access before the
failure occurred.
458 IBM System Storage DS8000 Series: Copy Services in Open Environments
as we will see in “Step 7: Start Metro Mirror from intermediate to local site” on page 462.
Example 31-11 shows the command to enable incremental re-synchronization at the
intermediate site. To verify, the lssprc -l command is issued.
In addition, to stop data transfer to the remote site, the Global Copy relationship from the
intermediate to the remote site is suspended. By suspending the Global Copy relationship,
the intermediate and remote sites’ volumes will both be in a suspended state and the change
recording bitmaps will remain active at the intermediate site.
In Example 31-12 the volumes of a certain application host are removed from the session. To
ensure that a final consistency group is formed, the Global Mirror is paused first. When the
volumes have been removed from the session the Global Mirror is resumed. Finally, the
Global Copy intermediate to remote is suspended.
Example 31-12 Removing volumes from the session and suspending the Global Copy intermediate to remote
dscli> pausegmir -dev IBM.2107-1301261 -lss 20 -session ad
CMUC00163I pausegmir: Global Mirror for session ad successfully paused.
dscli>
dscli> chsession -dev IBM.2107-1301261 -lss 20 -action remove -volume 2000-2007 ad
CMUC00147I chsession: Session ad successfully modified.
dscli>
dscli> resumegmir -dev IBM.2107-1301261 -lss 20 -session ad
CMUC00164I resumegmir: Global Mirror for session ad successfully resumed.
dscli>
dscli> lssession -dev IBM.2107-1301261 20 ad
LSS ID Session Status Volume VolumeStatus PrimaryStatus SecondaryStatus FirstPassComplete AllowCascading
========================================================================================================
20 AD - - - - - - -
Note: Remote to local site out-of-sync tracks must be drained completely before continuing
to the next step. This will ensure that the local site has been fully re-synchronized.
In Example 31-13 the command to suspend the Global Copy from remote to local is shown.
460 IBM System Storage DS8000 Series: Copy Services in Open Environments
This step is necessary to allow the failback of the intermediate to the remote site when
reversing Global Copy in “Step 6: Reverse Global Copy to run local to remote site” on
page 461.
Example 31-14 shows how to terminate the Global copy at the remote site.
The local to remote failover is executed at the local site and using cascading Global Copy
mode options. The failover will change the local site volumes status to become primary
suspended. Next the failback Global Copy local to remote site is executed at the local site with
cascading allowed and mode Global Copy.
A successful failover and failback reverses the Global Copy, which now runs from the local to
the remote site. Global Copy can be established to run from the local to the remote site with
Global Copy mode and cascading allowed.
Example 31-15 shows how to reverse the direction of the Global Copy between local and
remote.
Example 31-15 Reverse the direction of the Global Copy between local to remote
dscli> failoverpprc -dev IBM.2107-1301651 -remotedev IBM.2107-1300561 -type gcp -cascade
2000-2007:2000-2007
CMUC00196I failoverpprc: Remote Mirror and Copy pair 2000:2000 successfully reversed.
CMUC00196I failoverpprc: Remote Mirror and Copy pair 2001:2001 successfully reversed.
CMUC00196I failoverpprc: Remote Mirror and Copy pair 2002:2002 successfully reversed.
CMUC00196I failoverpprc: Remote Mirror and Copy pair 2003:2003 successfully reversed.
CMUC00196I failoverpprc: Remote Mirror and Copy pair 2004:2004 successfully reversed.
CMUC00196I failoverpprc: Remote Mirror and Copy pair 2005:2005 successfully reversed.
CMUC00196I failoverpprc: Remote Mirror and Copy pair 2006:2006 successfully reversed.
CMUC00196I failoverpprc: Remote Mirror and Copy pair 2007:2007 successfully reversed.
dscli>
dscli> failbackpprc -dev IBM.2107-1301651 -remotedev IBM.2107-1300561 -type gcp -cascade
2000-2007:2000-2007
CMUC00197I failbackpprc: Remote Mirror and Copy pair 2000:2000 successfully failed back.
CMUC00197I failbackpprc: Remote Mirror and Copy pair 2001:2001 successfully failed back.
CMUC00197I failbackpprc: Remote Mirror and Copy pair 2002:2002 successfully failed back.
CMUC00197I failbackpprc: Remote Mirror and Copy pair 2003:2003 successfully failed back.
CMUC00197I failbackpprc: Remote Mirror and Copy pair 2004:2004 successfully failed back.
CMUC00197I failbackpprc: Remote Mirror and Copy pair 2005:2005 successfully failed back.
Note: At this point Metro Mirror will need to become full duplex and Global Copy will need
to complete a first pass before going on to the next step.
In Example 31-16 the Metro Mirror from intermediate to local is established with the option
-incrementalresync override, which causes the copying of all tracks recorded in the change
recording bitmap at the intermediate site. This can be monitored with the lspprc -l command
shown next in the example. When all tracks have been drained, the incremental
re-synchronization is enabled with the option -incrementalresync enable.
Example 31-16 Start Metro Mirror from intermediate to local with incremental re-synchronization
dscli> mkpprc -dev IBM.2107-1301261 -remotedev IBM.2107-1301651 -type mmir -mode nocp -incrementalresync override 2000-2007:2000-2007
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2000:2000 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2001:2001 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2002:2002 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2003:2003 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2004:2004 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2005:2005 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2006:2006 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2007:2007 successfully created.
dsclI>
lspprc -dev IBM.2107-1301651 -l 2000-2007
ID State Reason Type Out Of Sync Tracks Tgt Read Src Cascade Tgt Cascade Date Suspended SourceLSS Timeout (secs) Critical
Mode First Pass Status Incremental Resync Tgt Write GMIR CG PPRC CG
=====================================================================================================================================================
======================================================================
2000:2000 Target Full Duplex - Metro Mirror 0 Disabled Invalid Enabled - 20 unknown Disabled
Invalid Disabled Disabled Disabled Disabled
2000:2000 Copy Pending - Global Copy 9 Disabled Enabled Invalid - 20 unknown Disabled
True Disabled Disabled Disabled Disabled
2001:2001 Target Full Duplex - Metro Mirror 0 Disabled Invalid Enabled - 20 unknown Disabled
Invalid Disabled Disabled Disabled Disabled
2001:2001 Copy Pending - Global Copy 7 Disabled Enabled Invalid - 20 unknown Disabled
True Disabled Disabled Disabled Disabled
2002:2002 Target Full Duplex - Metro Mirror 0 Disabled Invalid Enabled - 20 unknown Disabled
Invalid Disabled Disabled Disabled Disabled
2002:2002 Copy Pending - Global Copy 9 Disabled Enabled Invalid - 20 unknown Disabled
True Disabled Disabled Disabled Disabled
2003:2003 Target Full Duplex - Metro Mirror 0 Disabled Invalid Enabled - 20 unknown Disabled
Invalid Disabled Disabled Disabled Disabled
2003:2003 Copy Pending - Global Copy 7 Disabled Enabled Invalid - 20 unknown Disabled
True Disabled Disabled Disabled Disabled
2004:2004 Target Full Duplex - Metro Mirror 0 Disabled Invalid Enabled - 20 unknown Disabled
Invalid Disabled Disabled Disabled Disabled
2004:2004 Copy Pending - Global Copy 8 Disabled Enabled Invalid - 20 unknown Disabled
True Disabled Disabled Disabled Disabled
2005:2005 Target Full Duplex - Metro Mirror 0 Disabled Invalid Enabled - 20 unknown Disabled
Invalid Disabled Disabled Disabled Disabled
2005:2005 Copy Pending - Global Copy 11 Disabled Enabled Invalid - 20 unknown Disabled
True Disabled Disabled Disabled Disabled
2006:2006 Target Full Duplex - Metro Mirror 0 Disabled Invalid Enabled - 20 unknown Disabled
Invalid Disabled Disabled Disabled Disabled
2006:2006 Copy Pending - Global Copy 6 Disabled Enabled Invalid - 20 unknown Disabled
True Disabled Disabled Disabled Disabled
2007:2007 Target Full Duplex - Metro Mirror 0 Disabled Invalid Enabled - 20 unknown Disabled
Invalid Disabled Disabled Disabled Disabled
2007:2007 Copy Pending - Global Copy 14 Disabled Enabled Invalid - 20 unknown Disabled
True Disabled Disabled Disabled Disabled
462 IBM System Storage DS8000 Series: Copy Services in Open Environments
dsclI>
#
# Wait until all Out Of Sync Tracks has drained
#
dscli> mkpprc -dev IBM.2107-1301261 -remotedev IBM.2107-1301651 -type mmir -mode nocp -incrementalresync enable 2000-2007:2000-2007
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2000:2000 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2001:2001 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2002:2002 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2003:2003 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2004:2004 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2005:2005 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2006:2006 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2007:2007 successfully created.
dsclI>
dscli> lspprc -dev IBM.2107-1301261 -l 2000-2007
ID State Reason Type Out Of Sync Tracks Tgt Read Src Cascade Tgt Cascade Date Suspended SourceLSS Timeout (secs) Critical Mode F
irst Pass Status Incremental Resync Tgt Write GMIR CG PPRC CG
=====================================================================================================================================================
===============================================================
2000:2000 Full Duplex - Metro Mirror 0 Disabled Enabled Invalid - 20 120 Disabled I
nvalid Enabled Disabled Disabled Disabled
2001:2001 Full Duplex - Metro Mirror 0 Disabled Enabled Invalid - 20 120 Disabled I
nvalid Enabled Disabled Disabled Disabled
2002:2002 Full Duplex - Metro Mirror 0 Disabled Enabled Invalid - 20 120 Disabled I
nvalid Enabled Disabled Disabled Disabled
2003:2003 Full Duplex - Metro Mirror 0 Disabled Enabled Invalid - 20 120 Disabled I
nvalid Enabled Disabled Disabled Disabled
2004:2004 Full Duplex - Metro Mirror 0 Disabled Enabled Invalid - 20 120 Disabled I
nvalid Enabled Disabled Disabled Disabled
2005:2005 Full Duplex - Metro Mirror 0 Disabled Enabled Invalid - 20 120 Disabled I
nvalid Enabled Disabled Disabled Disabled
2006:2006 Full Duplex - Metro Mirror 0 Disabled Enabled Invalid - 20 120 Disabled I
nvalid Enabled Disabled Disabled Disabled
2007:2007 Full Duplex - Metro Mirror 0 Disabled Enabled Invalid - 20 120 Disabled I
nvalid Enabled Disabled Disabled Disabled
dscli>
dscli>
Note: In some cases, production may remain running at the intermediate site with a
three-site Metro/Global Mirror disaster recovery solution in place. In this case, no further
steps are taken.
In other cases, it may be necessary to restore production at the local site. In that case,
proceed to 31.3.3, “Returning to the original configuration” on page 463.
Primary
production
host Local Site
5
Remote Site
9
7
8 A
4
2 12
C
Intermediate Site 11
10
6 D
´Secondary 3
production
host B
1
13
Figure 31-7 Move incremental Resync Metro/Global Mirror back to local site
The following steps move production back to the local site and restore Incremental Resync
Metro/Global Mirror to its original configuration:
1. Stop applications at the intermediate site.
2. Suspend Metro Mirror from the intermediate site to the local site.
3. Fail over Global Copy from the remote to the intermediate site.
4. Terminate Metro Mirror at the local site.
5. Start applications at the local site.
6. Fail back Global Copy remote to the intermediate site.
7. Start Incremental Resync at the local site.
8. Terminate Global Mirror at the local site.
9. Suspend and terminate Global Copy from the local to the remote site.
10.Suspend Global Copy from the remote to the intermediate site.
11.Reverse Global Copy to run from the intermediate to the remote site.
12.Establish Metro Mirror at the local to the remote site.
13.Start Global Mirror at the intermediate site.
464 IBM System Storage DS8000 Series: Copy Services in Open Environments
Example 31-17 shows the DSCLI command to suspend the Metro Mirror.
Example 31-18 Fail over Global Copy from remote to intermediate site
dscli> failoverpprc -dev IBM.2107-1300561 -remotedev IBM.2107-1301261 -type gcp -cascade -force
2000-2007:2000-2007
CMUC00196I failoverpprc: Remote Mirror and Copy pair 2000:2000 successfully reversed.
CMUC00196I failoverpprc: Remote Mirror and Copy pair 2001:2001 successfully reversed.
CMUC00196I failoverpprc: Remote Mirror and Copy pair 2002:2002 successfully reversed.
CMUC00196I failoverpprc: Remote Mirror and Copy pair 2003:2003 successfully reversed.
CMUC00196I failoverpprc: Remote Mirror and Copy pair 2004:2004 successfully reversed.
CMUC00196I failoverpprc: Remote Mirror and Copy pair 2005:2005 successfully reversed.
CMUC00196I failoverpprc: Remote Mirror and Copy pair 2006:2006 successfully reversed.
CMUC00196I failoverpprc: Remote Mirror and Copy pair 2007:2007 successfully reversed.
dscli>
dscli>
Note: The writes from the remote to the intermediate site must be completely drained or
have completed the first pass before continuing to the next step. All changes at the local
site should have already been updated to the remote site, which is also being updated to
the intermediate site with the failover with force.
Example 31-20 shows how to issue the failbackpprc command with -force option. To monitor
the Out Of Sync Tracks a lspprc -l command must be issued at the remote site. The output
shows that the Global Copy from remote to intermediate is still a cascaded relation of the
Global Copy from local to remote.
466 IBM System Storage DS8000 Series: Copy Services in Open Environments
2007:2007 Target Copy Pending - Global Copy 0 Disabled Invalid Enabled - 20 unknown Disabled
Invalid Disabled Disabled Disabled Disabled
2007:2007 Copy Pending - Global Copy 0 Disabled Enabled Invalid - 20 unknown Disabled
True Disabled Disabled Disabled Disabled
dscli>
dscli>lspprc -dev IBM.2107-1300561 -fullid 2000-2007
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
=====================================================================================================================================================
IBM.2107-1300561/2000:IBM.2107-1301261/2000 Copy Pending - Global Copy IBM.2107-1300561/20 unknown Disabled True
IBM.2107-1300561/2001:IBM.2107-1301261/2001 Copy Pending - Global Copy IBM.2107-1300561/20 unknown Disabled True
IBM.2107-1300561/2002:IBM.2107-1301261/2002 Copy Pending - Global Copy IBM.2107-1300561/20 unknown Disabled True
IBM.2107-1300561/2003:IBM.2107-1301261/2003 Copy Pending - Global Copy IBM.2107-1300561/20 unknown Disabled True
IBM.2107-1300561/2004:IBM.2107-1301261/2004 Copy Pending - Global Copy IBM.2107-1300561/20 unknown Disabled True
IBM.2107-1300561/2005:IBM.2107-1301261/2005 Copy Pending - Global Copy IBM.2107-1300561/20 unknown Disabled True
IBM.2107-1300561/2006:IBM.2107-1301261/2006 Copy Pending - Global Copy IBM.2107-1300561/20 unknown Disabled True
IBM.2107-1300561/2007:IBM.2107-1301261/2007 Copy Pending - Global Copy IBM.2107-1300561/20 unknown Disabled True
IBM.2107-1301651/2000:IBM.2107-1300561/2000 Target Copy Pending - Global Copy IBM.2107-1301651/20 unknown Disabled Invalid
IBM.2107-1301651/2001:IBM.2107-1300561/2001 Target Copy Pending - Global Copy IBM.2107-1301651/20 unknown Disabled Invalid
IBM.2107-1301651/2002:IBM.2107-1300561/2002 Target Copy Pending - Global Copy IBM.2107-1301651/20 unknown Disabled Invalid
IBM.2107-1301651/2003:IBM.2107-1300561/2003 Target Copy Pending - Global Copy IBM.2107-1301651/20 unknown Disabled Invalid
IBM.2107-1301651/2004:IBM.2107-1300561/2004 Target Copy Pending - Global Copy IBM.2107-1301651/20 unknown Disabled Invalid
IBM.2107-1301651/2005:IBM.2107-1300561/2005 Target Copy Pending - Global Copy IBM.2107-1301651/20 unknown Disabled Invalid
IBM.2107-1301651/2006:IBM.2107-1300561/2006 Target Copy Pending - Global Copy IBM.2107-1301651/20 unknown Disabled Invalid
IBM.2107-1301651/2007:IBM.2107-1300561/2007 Target Copy Pending - Global Copy IBM.2107-1301651/20 unknown Disabled Invalid
dscli>
dscli>
In Example 31-23 the Global Copy from the local to the remote site is suspended.
Subsequently, the Global Copy from remote to intermediate is queried to check that all Out Of
Sync Tracks has been drained. If this is the case the Global Copy relation at the remote site is
removed.
Example 31-23 Suspend and remove Global Copy local to remote at remote
dscli> pausepprc -dev IBM.2107-1301651 -remotedev IBM.2107-1300561 2000-2007:2000-2007
CMUC00157I pausepprc: Remote Mirror and Copy volume pair 2000:2000 relationship successfully paused.
CMUC00157I pausepprc: Remote Mirror and Copy volume pair 2001:2001 relationship successfully paused.
CMUC00157I pausepprc: Remote Mirror and Copy volume pair 2002:2002 relationship successfully paused.
CMUC00157I pausepprc: Remote Mirror and Copy volume pair 2003:2003 relationship successfully paused.
CMUC00157I pausepprc: Remote Mirror and Copy volume pair 2004:2004 relationship successfully paused.
CMUC00157I pausepprc: Remote Mirror and Copy volume pair 2005:2005 relationship successfully paused.
CMUC00157I pausepprc: Remote Mirror and Copy volume pair 2006:2006 relationship successfully paused.
CMUC00157I pausepprc: Remote Mirror and Copy volume pair 2007:2007 relationship successfully paused.
dscli>
#
# Wait until all OOS has drained from remote to intermediate
#
468 IBM System Storage DS8000 Series: Copy Services in Open Environments
CMUC00155I rmpprc: Remote Mirror and Copy volume pair :2003 relationship successfully withdrawn.
CMUC00155I rmpprc: Remote Mirror and Copy volume pair :2004 relationship successfully withdrawn.
CMUC00155I rmpprc: Remote Mirror and Copy volume pair :2005 relationship successfully withdrawn.
CMUC00155I rmpprc: Remote Mirror and Copy volume pair :2006 relationship successfully withdrawn.
CMUC00155I rmpprc: Remote Mirror and Copy volume pair :2007 relationship successfully withdrawn.
dscli>
dscli>
A failover of Global Copy is first executed with cascading allowed from intermediate to remote
site. The failover will cause the intermediate site volumes status to be primary suspended and
sets up for local to intermediate and intermediate to remote connection. Next the failback
Global Copy intermediate to remote site is executed at the intermediate site with cascading
allowed and mode Global Copy.
The result of the failover and failback has successfully reversed the Global Copy relationship
to run from the intermediate to the remote site. Global Copy can now be established from the
intermediate to the remote site with Global Copy mode and cascading options.
The forced mirror also causes the change recording bitmaps that were created in “Step 7:
Start incremental re-synchronization at local site” on page 467 to be merged with the
out-of-sync bitmaps at the local site. Then Metro Mirror is established with incremental resync
initialized, which will create new change recording bitmaps for Metro Mirror at the local site.
Note: At this point Metro Mirror will need to become full duplex and Global Copy will need
to complete first pass before going on to the next step.
Tip: In case that at the failure of the local site (see 31.3.1, “Local site fails” on page 455) a
freezepprc command has been issued, the paths from local to intermediate have been
removed. In order to succeed with the failback from local to intermediate the paths must be
reestablished right now.
Example 31-26 shows that the option -incrementalresync override is used first to copy the
track marked in the change recording bitmap at the local site. All Out Of Sync Tracks have
been drained to the intermediate site. When done, the incremental re-synchronization is
started with empty bitmaps at the local site.
470 IBM System Storage DS8000 Series: Copy Services in Open Environments
dscli> mkpprc -dev IBM.2107-1301651 -remotedev IBM.2107-1301261 -type mmir -mode nocp -incrementalresync enable 2000-2007:2000-2007
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2000:2000 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2001:2001 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2002:2002 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2003:2003 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2004:2004 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2005:2005 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2006:2006 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2007:2007 successfully created.
dscli>
dscli> lspprc -dev IBM.2107-1301651 -l 2000-2007
ID State Reason Type Out Of Sync Tracks Tgt Read Src Cascade Tgt Cascade Date Suspended SourceLSS Timeout (secs) Critical Mode F
irst Pass Status Incremental Resync Tgt Write GMIR CG PPRC CG
=====================================================================================================================================================
==============================================================
2000:2000 Full Duplex - Metro Mirror 0 Disabled Enabled Invalid - 20 120 Disabled I
nvalid Enabled Disabled Disabled Enabled
2001:2001 Full Duplex - Metro Mirror 0 Disabled Enabled Invalid - 20 120 Disabled I
nvalid Enabled Disabled Disabled Enabled
2002:2002 Full Duplex - Metro Mirror 0 Disabled Enabled Invalid - 20 120 Disabled I
nvalid Enabled Disabled Disabled Enabled
2003:2003 Full Duplex - Metro Mirror 0 Disabled Enabled Invalid - 20 120 Disabled I
nvalid Enabled Disabled Disabled Enabled
2004:2004 Full Duplex - Metro Mirror 0 Disabled Enabled Invalid - 20 120 Disabled I
nvalid Enabled Disabled Disabled Enabled
2005:2005 Full Duplex - Metro Mirror 0 Disabled Enabled Invalid - 20 120 Disabled I
nvalid Enabled Disabled Disabled Enabled
2006:2006 Full Duplex - Metro Mirror 0 Disabled Enabled Invalid - 20 120 Disabled I
nvalid Enabled Disabled Disabled Enabled
2007:2007 Full Duplex - Metro Mirror 0 Disabled Enabled Invalid - 20 120 Disabled I
nvalid Enabled Disabled Disabled Enabled
dscli>
dscli>
Note: You can now verify that Consistency Groups are forming successfully.
In Example 31-27 we just add the volumes to the existing Global Mirror because in “Step 3:
Terminate Global Mirror and suspend Global Copy” on page 459 we remove the application
host’s volumes from the session instead of terminating the whole Global Mirror. The Global
Mirror is inspected if consistency formation is ongoing by two subsequent showgmir -metrics
commands to see whether the Total Successful CG Count is increasing.
472 IBM System Storage DS8000 Series: Copy Services in Open Environments
When the intermediate is fully re-synchronized, the original configuration can be restored
and Global Mirror is started at the intermediate site.
When the storage at the intermediate site is no longer accessible (failure), data will be copied
from the local to the remote site using the incremental re-synchronization implementation in
place.
When transitioning from the original MGM configuration to the configuration of the local to the
remote site, consistency groups are formed at local and drained to remote, while data
consistency is retained between the two sites. Production will be able to continue at the local
site until the intermediate site is operational again. Figure 31-8 illustrates the steps to recover
from a failure at the intermediate site while production is able to continue at the local site.
Primary
production
host Local Site
Remote Site
5
A
6
1
C
Intermediate Site
2
D
4
The steps for recovery after failure at the intermediate site are:
1. Suspend Metro Mirror at local site.
2. Clean up surviving components of Global Mirror (if possible).
3. Fail over Global Copy at remote site.
4. Verify Global Mirror consistency group.
5. Start Global Copy from local to remote site.
6. Create sessions and restart Global Mirror at local site.
Example 31-28 shows the command to suspend the Metro Mirror at the local site.
Note: Depending on the extent of the failure at the intermediate, Global Mirror may no
longer be running, and may show “FATAL” or failing consistency groups. If Global Mirror
was in the middle of a FlashCopy, then the consistency group may need to be verified at
some point. To check the status of the Global Mirror and see if consistency groups are
failing, the command showgmir -metrics will list the current status.
The command issued at the remote site is a failoverpprc with the -cascade option, as
shown in Example 31-29.
474 IBM System Storage DS8000 Series: Copy Services in Open Environments
CMUC00196I failoverpprc: Remote Mirror and Copy pair 2005:2005 successfully reversed.
CMUC00196I failoverpprc: Remote Mirror and Copy pair 2006:2006 successfully reversed.
CMUC00196I failoverpprc: Remote Mirror and Copy pair 2007:2007 successfully reversed.
dscli>
dscli>
To verify the Global Mirror consistency, see 27.7, “Check consistency at the remote site” on
page 400 describes how to verify the consistency group and determine if any action needs to
be taken.
Tip: The lsflash -l command will show a query of the FlashCopy status. This will be
helpful when verifying the Global Mirror consistency group.
When the Global Copy with incremental resync is completely established, the function that
was running previously at the local site is stopped. The current change recording bitmaps at
the local site are left alone and merged with the out-of-sync bitmaps.
If there was a former incremental resync relationship at the remote site, then the override
parameter must be used with the -incrementalresync option when establishing Global Copy.
Important: All writes are transferred at this point from the local to the remote site and all
the out-of-sync tracks must have drained before continuing to the next step. To query
out-of-sync tracks, issue the command lspprc -l at the local site.
In Example 31-30 the lspprc command issued at the local site shows the relation of the
Metro Mirror from the local to the intermediate site. After the mkpprc with -incrementalresync
recover the pair relation has changed to the Global Copy relation local to remote.
Example 31-30 Start Global Copy from local to remote with incremental re-synchronization
dscli> lspprc -dev IBM.2107-1301651 -fullid 2000-2007
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
================================================================================================================================================================
IBM.2107-1301651/2000:IBM.2107-1301261/2000 Suspended Internal Conditions Target Metro Mirror IBM.2107-1301651/20 300 Disabled Invalid
IBM.2107-1301651/2001:IBM.2107-1301261/2001 Suspended Internal Conditions Target Metro Mirror IBM.2107-1301651/20 300 Disabled Invalid
IBM.2107-1301651/2002:IBM.2107-1301261/2002 Suspended Internal Conditions Target Metro Mirror IBM.2107-1301651/20 300 Disabled Invalid
IBM.2107-1301651/2003:IBM.2107-1301261/2003 Suspended Internal Conditions Target Metro Mirror IBM.2107-1301651/20 300 Disabled Invalid
IBM.2107-1301651/2004:IBM.2107-1301261/2004 Suspended Internal Conditions Target Metro Mirror IBM.2107-1301651/20 300 Disabled Invalid
IBM.2107-1301651/2005:IBM.2107-1301261/2005 Suspended Internal Conditions Target Metro Mirror IBM.2107-1301651/20 300 Disabled Invalid
IBM.2107-1301651/2006:IBM.2107-1301261/2006 Suspended Internal Conditions Target Metro Mirror IBM.2107-1301651/20 300 Disabled Invalid
IBM.2107-1301651/2007:IBM.2107-1301261/2007 Suspended Internal Conditions Target Metro Mirror IBM.2107-1301651/20 300 Disabled Invalid
dscli>
dsclI> mkpprc -dev IBM.2107-1301651 -remotedev IBM.2107-1300561 -type gcp -mode full -incrementalresync recover 2000-2007:2000-2007
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2000:2000 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2001:2001 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2002:2002 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2003:2003 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2004:2004 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2005:2005 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2006:2006 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2007:2007 successfully created.
The Global Mirror session can now be started by issuing the mkgmir command at the local
site. This configuration remains unchanged until the intermediate site is available again for
Metro/Global Mirror.
Production continues to run at the local site without interruption while the original
Metro/Global Mirror configuration of local to intermediate to remote site transitions to local to
remote site.
Example 31-31 shows the steps to create sessions and Global Mirror and how to check it.
Example 31-31 Create session and start Global Mirror at local site
dscli> mksession -dev IBM.2107-1301651 -lss 20 10
CMUC00145I mksession: Session 10 opened successfully.
dscli>
dscli> chsession -dev IBM.2107-1301651 -lss 20 -action add -volume 2000-2007 10
CMUC00147I chsession: Session 10 successfully modified.
dsclI>
dsciI> lssession -dev IBM.2107-1301651 20 10
LSS ID Session Status Volume VolumeStatus PrimaryStatus SecondaryStatus FirstPassComplete AllowCascading
=================================================================================================================
20 10 Normal 2000 Join Pending Primary Copy Pending Secondary Simplex True Disable
20 10 Normal 2001 Join Pending Primary Copy Pending Secondary Simplex True Disable
20 10 Normal 2002 Join Pending Primary Copy Pending Secondary Simplex True Disable
20 10 Normal 2003 Join Pending Primary Copy Pending Secondary Simplex True Disable
20 10 Normal 2004 Join Pending Primary Copy Pending Secondary Simplex True Disable
20 10 Normal 2005 Join Pending Primary Copy Pending Secondary Simplex True Disable
20 10 Normal 2006 Join Pending Primary Copy Pending Secondary Simplex True Disable
20 10 Normal 2007 Join Pending Primary Copy Pending Secondary Simplex True Disable
dscli>
dscli> mkgmir -dev IBM.2107-1301651 -lss 20 -session 10
CMUC00162I mkgmir: Global Mirror for session 10 successfully started.
dscli>
dscli> showgmir -metrics 20
ID IBM.2107-1301651/20
Total Failed CG Count 1
Total Successful CG Count 39
Successful CG Percentage 97
Failed CG after Last Success 0
476 IBM System Storage DS8000 Series: Copy Services in Open Environments
Last Successful CG Form Time 11/07/2006 13:46:01 MST
Coord. Time (seconds) 50
Interval Time (seconds) 0
Max Drain Time (seconds) 30
First Failure Control Unit -
First Failure LSS -
First Failure Status No Error
First Failure Reason -
First Failure Master State -
Last Failure Control Unit -
Last Failure LSS -
Last Failure Status No Error
Last Failure Reason -
Last Failure Master State -
Previous Failure Control Unit -
Previous Failure LSS -
Previous Failure Status No Error
Previous Failure Reason -
Previous Failure Master State -
dscli>
dscli> lssession -dev IBM.2107-1301651 20 10
LSS ID Session Status Volume VolumeStatus PrimaryStatus SecondaryStatus FirstPassComplete AllowCascading
=========================================================================================================================
20 10 CG In Progress 2000 Active Primary Copy Pending Secondary Simplex True Disable
20 10 CG In Progress 2001 Active Primary Copy Pending Secondary Simplex True Disable
20 10 CG In Progress 2002 Active Primary Copy Pending Secondary Simplex True Disable
20 10 CG In Progress 2003 Active Primary Copy Pending Secondary Simplex True Disable
20 10 CG In Progress 2004 Active Primary Copy Pending Secondary Simplex True Disable
20 10 CG In Progress 2005 Active Primary Copy Pending Secondary Simplex True Disable
20 10 CG In Progress 2006 Active Primary Copy Pending Secondary Simplex True Disable
20 10 CG In Progress 2007 Active Primary Copy Pending Secondary Simplex True Disable
dscli>
dscli>
There may be Metro Mirror or Global Mirror relationships or both that need to be terminated.
These relationships, depending on how the intermediate was lost, may not have been
cleaned up prior to the failure of the intermediate in Section “Step 2: Clean up surviving
components of Global Mirror (if possible)” on page 474. The components that were not
terminated at that stage must be terminated now.
Local Site
Remote Site
Primary
A
production
host
C
Intermediate Site
1 2 3
The steps to cleanup the intermediate after it is fully recovered are as follows:
1. Remove Metro Mirror.
2. Suspend Global Copy intermediate to remote at intermediate.
3. Terminate former Global Mirror.
The command rmpprc issued at the intermediate site will terminate all Metro Mirror
relationships at the intermediate.
Tip: To remove the Metro Mirror pairs, the communication between local and intermediate
site must work. We recommend that you check the PPRC paths in both directions when the
intermediate site became available again.
478 IBM System Storage DS8000 Series: Copy Services in Open Environments
In Example 31-32 the paths between local and intermediate are checked first. When the
Metro Mirror was removed at the local site the Global Copy between local and remote is show
with the subsequent lssprc command issued at the local site.
The command pausepprc issued at the intermediate site will suspend the Global Copy
relations between the intermediate and the remote site. The -unconditional -at src option must
be used with the pausepprc command, because Global Copy needs only to be suspended at
the intermediate, the Global Copy primary. The previous Global Copy secondary at the
remote site is already in a failover state, and thus is primary suspended.
Example 31-33 shows how to suspend Global Copy at the intermediate site.
It if was not terminated before and fails again now, it could be caused by orphaned
subordinates. In this case, Global Mirror must be terminated at the orphaned subordinate by
using the command rmgmir.
Note: The failover state at the remote site will prevent any previous Global Mirror
configuration from operating.
Figure 31-10 illustrates the steps to re-synchronize the intermediate before restoring the
original configuration.
Primary
production
host Local Site
Remote Site
2
A
C
Intermediate Site
D
1
480 IBM System Storage DS8000 Series: Copy Services in Open Environments
Step 1: Fail back Global Copy remote to intermediate site
At “Step 2: Clean up surviving components of Global Mirror (if possible)” on page 474, the
remote site volumes were changed to primary volumes suspended in preparation for the
failback to the remote site.
We can now issue a failbackpprc command at the remote site, and this begins copying data
from the remote site to the intermediate site.
Note: Waiting for the initial pass of the re-synchronization before restarting incremental
re-synchronization is good practice to reduce the number of updates sent later when Metro
Mirror is started with incremental re-synchronization and force at the local site. To query
the out-of-sync status, the lspprc -l command is issued at the intermediate site.
Example 31-34 shows the command to fail back the Global Copy.
The command mkpprc with the option -incrementalresync enablenoinit will successfully start
Incremental Resync at the local site without enabling the bitmaps.
Note: This step is necessary to allow the Metro Mirror relationship at the local site to be
restored in a later step using the -incrementalresync override option.
Primary
production Local Site
host
Remote Site
2
A 3
1
C
Intermediate Site 5
D
4
7 B
Note: The swap back to the intermediate can be done at any time but would normally be
done at a planned time.
482 IBM System Storage DS8000 Series: Copy Services in Open Environments
Example 31-36 Stop Global Mirror at local site
dscli> rmgmir -dev IBM.2107-1301651 -quiet -lss 20 -session 10
CMUC00165I rmgmir: Global Mirror for session 10 successfully stopped.
dscli> chsession -dev IBM.2107-1301651 -lss 20 -action remove -volume 2000-2007 10
CMUC00147I chsession: Session 10 successfully modified.
dscli> rmsession -dev IBM.2107-1301651 -quiet -lss 20 10
CMUC00146I rmsession: Session 10 closed successfully.
dscli>
dscli>
Important: Out-of-sync tracks need to be drained completely to the remote. Wait until this
is the case before continuing to the next step. To query the out-of-sync tracks, the lspprc
-l command can be issued from the remote site.
Example 31-37 shows the DSCLI command to suspend the Global Copy.
This step is necessary to allow the fail back of the intermediate to remote site in a later step.
Note: The local site has updates for the intermediate and remote being recorded in the
incremental resync change recording and out-of-sync bitmap.
#
# At Local Site
#
The command issued is failoverpprc at the intermediate. Because the volumes at the
intermediate are cascaded, the -cascade option is issued with the failoverpprc command.
This will make the intermediate primary suspended.
Note: In “Step 2: Suspend Global Copy at local to remote” on page 483 Global Copy was
suspended at the local, thus stopping any new data from going to the remote. At this point,
out-of-sync tracks should be drained. To check the out-of-sync tracks, issue the lspprc -l
command at the remote site.
Example 31-39 shows the appropriate DSCLI command to fail over the Global Copy.
484 IBM System Storage DS8000 Series: Copy Services in Open Environments
CMUC00196I failoverpprc: Remote Mirror and Copy pair 2007:2007 successfully reversed.
dscli>
dscli>
Next, to monitor and track data as it is being written on the primary volumes at the local site, a
Metro Mirror relationship with incremental resync is created from the local to the intermediate
site by using the mkpprc command with the -incrementalresync enable option. By using the
enable parameter for -incrementalresync, incremental resync is initialized by creating a new
change recording bitmap on the local site.
Note: Both Metro Mirror and Global Copy may still be in first pass. To query the status of
Global Mirror or Metro Mirror, use the lspprc -l command at the local site to query Metro
Mirror and at the intermediate site to query Global Copy.
Example 31-41 shows the DSCLI command to create the Metro Mirror with -incremental
override. During this phase all tracks marked in the incremental re-synchronization are not
enabled. When all tracks have been drained to the intermediate site the incremental
re-synchronization is enabled.
Example 31-41 Create Metro Mirror with incremental resync at local site
dscli> mkpprc -dev IBM.2107-1301651 -remotedev IBM.2107-1301261 -type mmir -mode nocp -incrementalresync override 2000-2007:2000-2007
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2000:2000 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2001:2001 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2002:2002 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 2003:2003 successfully created.
Important: Verify that the out-of-sync tracks have completed for Metro Mirror before
starting Global Mirror. Otherwise, the consistency groups will start to fail when Global
Mirror is started with the mkgmir command.
486 IBM System Storage DS8000 Series: Copy Services in Open Environments
Example 31-42 shows how to start the Global Mirror and the sessions at the intermediate
site.
Part 8 Interoperability
In this part we discuss the interoperability of the various Copy Services functions on the
DS8000, and also the interoperability of the DS8000 with other IBM System Storage and
TotalStorage disk subsystems in Copy Services implementations.
There are two possibilities for accomplishing data consistency during a migration using an
environment that has double cascading. The first possibility is to shut down all applications at
the local site and let the out-of-sync drain completely. The second possibility is to change the
Global Copy relationships to Metro Mirror, then when the out-of-sync is approaching zero, a
freeze is issued to the changed relationships.
DWDM DWDM
Figure 32-1 shows double cascading with a Metro Mirror relationship from the local to the
secondary and Global Copy from the secondary to the tertiary site and from the tertiary to the
remote site. There is a cascading relationship at both the secondary and tertiary sites where
the volumes are both secondaries and primaries. In this example, if all applications are
stopped at the local site, then the local, secondary, tertiary and remote will reach a point after
some time where they are all equal. Therefore they will be consistent and data has
successfully been migrated to the remote site volumes.
The second approach to provide data consistency during migration, again looking at the
example in Figure 32-1, a freeze is first issued to the Metro Mirror relationship from the local
to the secondary site. Since Metro Mirror is running between the local and the secondary site,
there is already a synchronous relationship. Therefore the secondary volumes are consistent
with the local site volumes at this time. The next step would be to change all the Global Copy
relationships to Metro Mirror relations and then issue a freeze when the out-of-sync has
reached zero. When the out-of-sync is fully drained, then there is a consistent relationship
from the secondary to the remote and migration to the remote is complete. Data migration
has sent a snapshot of data to the remote site so that there is consistency. This process
would be used when production cannot be stopped, since production remains to be running
at the local site. However, once the freeze is first issued on the Metro Mirror relationship, from
that point forward the remote will only be consistent with the secondary. The local during this
492 IBM System Storage DS8000 Series: Copy Services in Open Environments
process is still being updated and is changing. The remote will have a snapshot of the time
since the freeze was issued.
DWDM DWDM
Once data migration and consistency have been accomplished at the remote site, you can
start production at the remote site, if desired. In this case all applications must first be
stopped at the local site and the direction of the mirror is reversed so that Metro Mirror is now
running from the remote to the tertiary site and Global Copy is running from the tertiary site to
the secondary site and then to the primary site. Figure 32-2 shows the new configuration with
the Metro Mirror and Global Copy reversed and production running at the remote site.
In this example, it is shown that to protect the production at the remote, the remote copy
relations can be reversed to use the tertiary, secondary, or local volumes as targets. These
copy relations would be created before starting production at the remote site. Since the
relationships from the secondary to the tertiary and from the tertiary to the remote are both in
a cascaded relationship, to reverse the directions Global Copy must be first removed and
then recreated in the reverse directions.
Note: No I/O should be running at any time at any of the sites while the reversal of the
remote copy relations is being done. This would corrupt the data resulting and
consequently requiring a full copy.
Note: Remote Mirror and Copy (RMC) was formerly called Peer-to-Peer Remote Copy
(PPRC). All references to PPRC are interchangeable with RMC.
The DS6000 and DS8000 must also have Fibre Channel adapters that have connectivity to
each other, either directly or via Fibre Channel switches. ESCON adapters cannot be used to
mirror a DS6000 with a DS8000.
For creating Remote Mirror and Copy (RMC), also known as Peer-to-Peer Remote Copy
(PPRC) paths and pairs:
In a mixed environment, one machine will be the source machine (local) and one will be
the target machine (remote). If you want the ability to manage the RMC paths and pairs
from either machine, then you will need network connectivity between the DS6000 SMC
and the DS8000 HMC. If you use the DS CLI, the DS SMC and DS HMC do not need to
be able to communicate with each other, but the DS CLI machine needs to be able to
communicate with both the DS HMC and the DS SMC.
496 IBM System Storage DS8000 Series: Copy Services in Open Environments
If you do not plan to use the remote system as a source and you do not intend to use the
DS Storage Manager GUI to manage the pairs and paths, then network connectivity to the
remote machine (assuming the remote machine is at a remote site) is not necessary. This
is because, when you use the DS CLI, all path and pair establishment is done by
connecting to the source machine (the local machine). In a disaster recovery scenario,
you will have to travel to the remote site to perform management tasks for the remote
machine. This setup is not recommended because it is less flexible.
Note: A DS8000 Storage Complex is made of up of either one or two HMCs that together
manage one or two DS8000 Storage Units. A DS6000 Storage Complex consists of either
one or two SMCs that together manage one or two DS6000 Storage Units.
Placing the password in the profile is not very secure (because it is stored in a plain text file),
but may prove more convenient. A password file can be created using the managepwfile and
is a better way to manage this. After creating the password file (which by default is called
security.dat), you can remove the password from the profile and instead specify the pwfile
file.
A simple method when you have multiple servers to manage is to create multiple profile files.
Then, when starting the DS CLI, you can specify the profile file you wish to use with the -cfg
parameter. In a Windows environment you could have multiple Windows batch files (BAT
files), one for each machine you wish to manage. The profile shown in Example 33-1 on
page 497 could be saved in the C:\program files\ibm\dscli\profile directory as
1750source.profile. Then you could create a simple Windows BAT file with three lines, as
shown in Example 33-2.
We save the BAT file onto the Windows desktop and start it by double-clicking the icon. The
DS CLI will open and will start the specified profile. By creating a BAT file and profile for each
machine, we can simplify the process of starting the DS CLI. We can also specify the profile
when starting the DS CLI from inside a script, or when running the DS CLI in script mode.
Important: Make sure the user ID you use to log on to the DS8000 DS GUI also exists on
the DS6000 SMC and that it has the same password. If not, the operation to add the
Storage Complex will fail. You will need to always use this same user ID for multi-complex
management.
498 IBM System Storage DS8000 Series: Copy Services in Open Environments
Figure 33-1 shows the Add Storage Complex panel.
6. Enter the IP address of the DS6000 SMC into the Management console 1 IP address box.
If you have a second SMC, select the Define a second Management console box and
enter the second SMC address into the Management console 2 IP address box. Now
select OK.
7. When the add Storage Complex operation completes, the Storage Complex screen will
now show the DS6000 SMC as an additional Storage Complex.
8. Having added the DS6000 Storage Complex to the DS8000 Storage Complex, you are
now able to use the DS8000 DS GUI to create paths and Remote Mirror and Copy pairs,
where the DS6000 is the source device. You can also use the DS8000 DS GUI to manage
FlashCopy pairs on the DS6000.
Note: The steps to add the DS6000 Storage Complex to the DS8000 Storage Complex
cannot be performed using the DS CLI. This is by design. However, if you do not plan to
use the GUI, this will not be an issue.
Important: Make sure the user ID you use to log on to the DS6000 DS GUI also exists on
the DS8000 HMC and that it has the same password. If not, the operation to add the
Storage Complex will fail. You will need to always use this same user ID for multi-complex
management.
6. Enter the IP address of the DS8000 HMC in the Management console 1 IPfield. If you
have a second DS8000 HMC, select Define a second Management console and enter
the second HMC address in the Management console 2 IP field. Click OK.
7. When the add Storage Complex operation completes, the Storage Complex panel shows
the DS8000 HMC as an additional Storage Complex. In Figure 33-3, a DS6000 SMC, an
ESS 800 Copy Services Server, and a DS8000 HMC are all defined in one DS6000 DS
GUI.
10.0.0.1
10.0.0.2
10.0.0.3
8. Having added the DS8000 Storage Complex to the DS6000 Storage Complex, you are
now able to use the DS6000 DS GUI to create paths and Remote Mirror and Copy pairs,
where the DS8000 is the source device. You can also use the DS6000 DS GUI to manage
FlashCopy pairs on the DS8000.
Note: The steps to add the DS8000 Storage Complex to the DS6000 Storage Complex
cannot be performed using the DS CLI.
User ID management
There are two important considerations when managing multiple Storage Complexes from a
single DS GUI:
The user ID you use to add the alternative complex must exist on both complexes. You
must continue to use this user ID for all multi complex operations. If you log on to either
DS GUI with a user ID that does not exist on the other complex, the remote complex will
not appear in the Storage Complex screen.
500 IBM System Storage DS8000 Series: Copy Services in Open Environments
Even when you have successfully added one complex to another, new user IDs created in
one complex will not mirror to the other complex. They have to be manually created and
managed in each complex.
Storage management
You cannot use the DS8000 DS GUI to perform storage configuration on a DS6000.
Likewise, you cannot use the DS6000 DS GUI to perform storage configuration on a DS8000.
You can only perform Copy Services management tasks on the alternative device. If you are
logged onto the DS6000 DS GUI and wish to configure some storage on the DS8000, you will
need to log off and then log on to the DS8000 DS GUI. The same applies if you are logged
onto the DS8000 DS GUI and wish to configure storage on the DS6000. You must log on to
the DS6000 DS GUI to do this.
When fixed block volumes are created on the DS8000, the user is given three size choices:
ds The number of bytes allocated will be the requested capacity value
times 230 .
ess The number of bytes allocated will be the requested capacity value
times 109 .
blocks The number of bytes allocated will be the requested capacity value
times 512 bytes (since each block is 512 bytes).
When creating volumes that are going to be either RMC source or RMC target between
DS6000 and DS8000, the recommended volume creation method is to use -type ds, as this
creates binary sized volumes that will not result in any wasted extent space. However, if the
planned partner device is an ESS 800, then you should only use -type ess or -type blocks.
Example 33-3 shows the resulting volumes. Each volume shows in the cap (2^30B) column
as being 2 GB and 4194304 blocks.
Where the size parameter in the formula is the decimal (10^9B) volume size. You could place
the output of lsfbvol into a spreadsheet and calculate the expected block count for each
volume using the contents of the 10^9 (decimal) cap column. Then compare the column that
has the output of the formula to the block column to ensure the two match. In Table 33-1 the
output of the lsfbvol command has been modified and placed into a spreadsheet. The
formula has been placed into the Expected Block Size column. It uses the values in the
(10^9B) column as an input to determine the correct block size. One volume, 6002, is 64
blocks larger than it should be. If that volume is to be used for RMC, it should be deleted and
re-created (provided the data on it can be moved).
502 IBM System Storage DS8000 Series: Copy Services in Open Environments
5100 FB 512 P3 - 4.5 8789120 8789120
Note: Do not apply this formula to System i volumes or volumes that only have a size in the
2^30 column. This is because System i volumes will be correctly sized using 520 byte
blocks, while volumes that are binary sized (their size is listed in the 2^30) will also have
the correct number of blocks.
In Example 33-4, a fixed block volume ID 1808 was created on the DS6800. We then
immediately attempted to use it as a RMC target. This attempt fails with the message, as
shown. Do not be concerned about the FlashCopy comment; this is normal. Wait for the
volume initialization process to finish and try again. The initialization time will vary, depending
on the size of the volume.
Example 33-4 Establishing a RMC (PPRC) pair where the target is still being formatted
dscli> mkpprc -remotedev IBM.1750-1300247 -type mmir 1406:1808
ate/Time: 3 November 2005 23:02:31 IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7503461
MUN03117E mkpprc: 1406:1808: Unable to establish FlashCopy or Remote Mirror and Copy pair.
A FlashCopy initialization is in progress.
For DS8000, port IDs look like I0000 or I0123, which actually breaks out as IEECP. The EE is
the enclosure number (minus 1), C is the card slot number (minus 1), and P is the port
number on the card. So I0123 is enclosure 2 (1+1), slot 3 (2+1), port 3.
For DS6000, port IDs range from I0000 to I0001 and I0100 to I0103, which actually breaks
out as IEECP. The EE is the controller number (00 or 01), C is the card slot number (always
zero), and P is the port number on the card (0 to 3). So I0103 is controller 1, card slot 0,
port 3.
33.3.3 Establish logical paths between DS8000 and DS6000 using DS CLI
You can also use the DS CLI to establish logical paths. One additional step is that you need
to know the WWNN of the target Storage Image. The three commands we will use to
establish a path are lssi, lsavailpprcport, and mkpprcpath.
504 IBM System Storage DS8000 Series: Copy Services in Open Environments
How to determine the DS8000 WWNN using DS GUI
To do this:
1. Click Real-time manager.
2. Click Manage hardware.
3. Click Storage images (not the Storage Unit).
4. Click in the Select column for the DS8000 Storage Image, and from the Select Action
pull-down, choose Properties.
5. The DS8000 Storage Image WWNN will be displayed on the subsequent properties
screen.
In Example 33-8, we first establish a single path between LSS 14 on the DS8000 and LSS 18
on the DS6000. We then display the paths available for LSS 14 on the DS8000.
In Example 33-9 we add an additional path between LSS 14 on the DS8000 and LSS 18 on
the DS6000. Take careful note though, we include the existing path when creating a new
path. Otherwise the existing path is removed and only the new path will be available for use.
To establish paths where the DS6000 is the source, we should connect to the DS6000 using
the DS CLI and follow the same process, but specify the DS8000 as the remote device.
Important: When you connect using the DS CLI to the DS8000 HMC, the DS8000 is the
local device and the DS6000 is the remove device. If you connect using the DS CLI to the
DS6000, then the DS6000 is now the local device and the DS8000 is the remote device.
Attention: The rmpprcpath command will remove all paths between the source and target
LSSs. To just reduce the path count, use the mkpprcpath command specifying only the
paths you wish to continue using.
33.4.1 Managing Metro Mirror or Global Copy pairs using the DS GUI
Establishing and managing Metro Mirror and Global Copy pairs between a DS6000 and a
DS8000 using the DS GUI is no different than using the DS GUI to establish pairs between
two DS8000s or DS6000s. See 14.4, “DS Storage Manager GUI” on page 136, and 22.3, “DS
Storage Manager GUI” on page 291.
506 IBM System Storage DS8000 Series: Copy Services in Open Environments
In Example 33-10 we create two pairs using the mkpprc command. We then list them using
the lspprc command. We then remove one pair using the rmpprc command.
In Example 33-10 we connected to the DS8000 HMC using the DS CLI, so the DS6000 is the
remote device. If we wish to establish pairs where the DS6000 is the source device, we need
to connect to the DS6000 using the DS CLI. This will make the DS8000 the remote device.
Important: If you specify the wrong -remotedev or you are using a profile where a different
remote device to the one you intended to work on is specified, you may get an error
message, CMUN03057, saying you are specifying an invalid subsystem ID. This may be
because you are specifying the wrong remote device serial number. If you have multiple
possible remote devices, do not place a remotedev entry in your DS CLI profile.
In Example 33-11 we create two pairs using the mkpprc command. We then list them using
the lspprc command. We then pause one pair using the pausepprc command.
508 IBM System Storage DS8000 Series: Copy Services in Open Environments
Fatal Reason Not Fatal
CG Interval (seconds) 0
XDC Interval(milliseconds) 50
CG Drain Time (seconds) 30
Current Time 11/03/2005 21:09:32 EST
CG Time -
Successful CG Percentage 0
FlashCopy Sequence Number 0x00000000
Master ID IBM.2107-7503461
Subordinate Count 0
Master/Subordinate Assoc -
Example 33-13 Creating a remote FlashCopy where the DS6000 is the remote target
dscli> lspprcpath 14
Date/Time: 3 November 2005 22:39:58 IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7503461
Src Tgt State SS Port Attached Port Tgt WWNN
=========================================================
14 18 Success FF18 I0001 I0103 500507630EFFFE16
14 18 Success FF18 I0131 I0003 500507630EFFFE16
dscli> mkpprc -remotedev IBM.1750-1300247 -type mmir 1405:1805
Date/Time: 3 November 2005 21:41:18 IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7503461
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 1405:1805 successfully created.
dscli> mkremoteflash -dev IBM.1750-1300247 -conduit IBM.2107-7503461/14 -persist -nocp 1805:1806
Date/Time: 3 November 2005 22:35:45 IBM DSCLI Version: 5.1.0.204 DS: IBM.1750-1300247
CMUC00173I mkremoteflash: Remote FlashCopy volume pair 1805:1806 successfully created. Use the
lsremoteflash command to determine copy completion.
dscli> lsremoteflash -dev IBM.1750-1300247 -conduit IBM.2107-7503461/14 1805:1806
Date/Time: 3 November 2005 22:36:00 IBM DSCLI Version: 5.1.0.204 DS: IBM.1750-1300247
ID SrcLSS SequenceNum ActiveCopy Recording Persistent Revertible SourceWriteEnabled
===========================================================================================================
1805:1806 18 0 Disabled Disabled Enabled Disabled Enabled
dscli> rmremoteflash -dev IBM.1750-1300247 -conduit IBM.2107-7503461/14 1805:1806
Date/Time: 3 November 2005 22:36:19 IBM DSCLI Version: 5.1.0.204 DS: IBM.1750-1300247
CMUC00179I rmremoteflash: Are you sure you want to remove the remote FlashCopy pair {0}? [y/n]:y
CMUC00180I rmremoteflash: Removal of the remote FlashCopy volume pair 1805:1806 has been initiated
successfully. Use the lsremoteflash command to determine when the relationship is deleted.
Note: Commands that work with remote FlashCopies use the -dev parameter to define the
machine on which the FlashCopy is to be performed. Other commands, such as mkpprc
commands, will refer to this same device as the remote device, using -remotedev.
However, since a FlashCopy has to be sent to the remote site and then performed locally
on that site, the use of the -dev parameter to refer to the remote machine is correct.
510 IBM System Storage DS8000 Series: Copy Services in Open Environments
34
Note: On the ESS 800, Remote Mirror and Copy (RMC) is called Peer-to-Peer Remote
Copy (PPRC). All references to PPRC are interchangeable with RMC.
The ESS 800 must also have Fibre Channel adapters that have connectivity to the DS8000.
ESCON adapters cannot be used to mirror an ESS 800 to a DS8000. You cannot have a
Copy Services relationship between a 2105 E20 or F20 and a DS8000. You could, however,
have a cascaded installation where an ESS F20 was mirrored to an ESS 800 that was then
mirrored to a DS8000. The ESS F20 to ESS 800 relationship would have to be managed with
ESS management tools such as the ESS CLI or the ESS Web Copy Services GUI.
For creating Remote Mirror and Copy (RMC), also known as Peer-to-Peer Remote Copy
(PPRC), paths and pairs:
If you wish to use the ESS 800 as a PPRC source machine, then you will need network
connectivity to the ESS 800 Copy Services servers, either from a machine running the DS
CLI, or from the DS8000 HMC.
If, however, the ESS 800 will purely be a remote target for PPRC, and you do not plan to
ever use it as a source machine and you do not intend to use the DS GUI to manage the
512 IBM System Storage DS8000 Series: Copy Services in Open Environments
pairs and paths, then you do not need to have network connectivity to the ESS 800 Copy
Services servers. This is because all path and pair establishment is done by connecting to
the source machine (which would be the DS8000). This setup is not recommended
because it is less flexible.
Important: The ESS Web Copy Services user IDs are not used by the DS CLI or DS GUI.
You only need to create a matching ESS Specialist user ID.
Putting the password in the profile is not very secure (because it is stored in plain text), but it
can be more convenient. A password file can be created using the managepwfile command
and is a better way to manage this. After creating the password file (which by default is called
security.dat), you can remove the password from the profile and instead specify the pwfile
file.
A simple method when you have multiple machines to manage is to create multiple profile
files. Then when starting the DS CLI, you can specify the profile file you wish to use with the
-cfg parameter. In a Windows environment you could have multiple Windows batch files (BAT
files), one for each machine you wish to manage. The profile shown in Example 34-1 on
page 513 could be saved in the C:\program files\ibm\dscli\profile directory as
2105source.profile. Then you create a simple Windows BAT file with three lines, as shown in
Example 34-2.
We save the BAT file onto the Windows desktop and start it by double-clicking the icon. The
DS CLI will open and will start the specified profile. By creating a BAT file and profile for each
machine, we can simplify the process of starting the DS CLI. We can also specify the profile
when starting the DS CLI from inside a script, or when running the DS CLI in script mode.
You add an ESS Copy Services domain with the DS Storage Manager to the DS8000 as
follows:
1. Click Real-time manager.
2. Click Manage hardware.
3. Click Storage complexes.
4. Select Add 2105 Copy Services Domain.
514 IBM System Storage DS8000 Series: Copy Services in Open Environments
Figure 34-1 shows the Add 2105 Copy Services Domain panel.
5. Enter the IP address of the ESS Copy Services server A in the Server 1 IP address box. If
you have a Server B, select the Define a second Copy Services server box and enter
the Server B IP address in the Server 2 IP address box. However, if Server B is running on
a 2105-F20, you should not define it.
Having added the ESS 800 Copy Services domain to the DS GUI, you are now able to use
the DS GUI to create paths and Remote Mirror and Copy pairs, where the ESS 800 is the
source device. You can also use the DS GUI to manage FlashCopy pairs on the ESS 800.
Note: The steps to add the ESS 800 Copy Services Domain to a DS8000 cannot be
performed using the DS CLI. However, if you do not plan to use the GUI, this will not be an
issue.
Restriction: You cannot use the ESS Copy Services Server Web GUI to manage Copy
Services relationships between an ESS 800 and a DS8000. The Volumes tab will only
show ESS 800 volumes, not DS8000 volumes. The Paths tab will not shows paths between
an ESS 800 and a DS8000. It will only show paths between ESS 800s. If you are using
PPRC between a DS8000 and an ESS 800, all management of that PPRC relationship
must be done with the DS CLI or via the DS GUI.
Storage management
You cannot use the DS8000 DS GUI to perform storage configuration on an ESS 800.
Likewise, you cannot use the ESS 800 Web Specialist to perform storage configuration on a
DS8000. You can only perform Copy Services management tasks on the alternative device. If
you are logged onto the DS8000 DS GUI and wish to configure some storage on the ESS
800, you will need to log on to the ESS 800 Web Specialist.
The correct method is to determine the ESS volume size and then create fixed block volumes
that are sized using the -type ess parameter.
Figure 34-2 Viewing ESS 800 volume size using ESS Specialist GUI
You can also use the ESS CLI to view the volume size, as shown in Example 34-3.
Example 34-3 View ESS 800 volume size using ESS CLI
C:\Program Files\ibm\ESScli>esscli -s 10.0.0.1 -u storwatch -p specialist list volume
Wed Nov 02 01:13:10 EST 2005 IBM ESSCLI 2.4.0
516 IBM System Storage DS8000 Series: Copy Services in Open Environments
Creating matching fixed block volumes on the DS8000
In Figure 34-2 on page 516 and Example 34-3 on page 516 we listed four volumes, which are
all shown as being 1.1 GB. In our example we are going to create volumes 1400 to 1403 on
the DS8000 in Extent Pool P0 to use as RMC (PPRC) target volumes. We can use the
following DS CLI command:
mkfbvol -extpool P0 -cap 1.1 -type ess 1400-1403
Example 34-4 shows the resulting volumes. Each volume shows in the cap (10^9B) column
as being 1.1 GB.
10.If the block counts do not match, then you must either:
– Remove the volume with rmfbvol and create it again (assuming you are now at bundle
6.0.500.31 or later).
– Create a new volume for the purposes of PPRC (again, assuming you are now at
bundle 6.0.500.31 or later).
– Use this volume only as a PPRC target, since the equivalent ESS 800 source volume
is slightly smaller.
Where the size parameter in the formula is the ESS 800 volume size. You could place the
output of lsfbvol into a spreadsheet and calculate the expected block size for each volume
using the contents of the 10^9 (decimal) cap column. Then compare the column that has the
output of the formula to the block column, to ensure the two match. In Table 34-1 the output of
the lsfbvol command has been modified and placed into a spreadsheet. The formula has
been placed into the Expected Block Size column. It uses the values in the (10^9B) column as
an input to determine the correct block size. One volume, 6002, is 64 blocks larger than it
should be. If that volume is to be used for PPRC, it should be deleted and re-created
(provided the data on it can be moved).
518 IBM System Storage DS8000 Series: Copy Services in Open Environments
5100 FB 512 P3 - 4.5 8789120 8789120
Note: Do not apply this formula to System i volumes or volumes that only have a size in the
2^30 column. This is because System i volumes will be correctly sized using 520 byte
blocks, while volumes that are binary sized (their size is listed in the 2^30) will also have
the correct number of blocks.
One option when creating fixed block volumes is to use the parameter -type blocks. The
number of blocks could be calculated using the methods shown in Figure 34-3 on page 518
or Table 34-1 on page 518. An example command would be:
mkfbvol -extpool P0 -type blocks -cap 10742208 1404
If we intend to use FlashCopy to copy ESS LUN 000-22399 onto 001-22399 using the DS
CLI, we must specify 1000 and 1001. If instead we specify 0000 and 0001, we will actually
run the FlashCopy against CKD volumes. This may result in an unplanned outage on the
System z environment that was using CKD volume 0001.
The ESS 800 port IDs do not follow the same rule. To decode them, take the last two digits in
the port ID and then use Figure 34-4. So port ID I0020 is the adapter in host bay 2, slot 1 and
port ID I00AC is the adapter in host bay 4, slot 4.
00 04 08 0C 20 24 28 2C 80 84 88 8C A0 A4 A8 AC
Slot 1 Slot 2 Slot 3 Slot 4 Slot 1 Slot 2 Slot 3 Slot 4 Slot 1 Slot 2 Slot 3 Slot 4 Slot 1 Slot 2 Slot 3 Slot 4
Host Bay 1 Host Bay 2 Host Bay 3 Host Bay 4
Tip: Open systems LSSs on an ESS 800 are always LSS 10 to LSS 1F. If you select ESS
800 LSS 00 to ESS 800 LSS 0F, you are working with System z LSSs on the ESS 800.
520 IBM System Storage DS8000 Series: Copy Services in Open Environments
5. Click Create; see Figure 34-5.
6. You will now be prompted to select the source LSS of the DS8000 from which you want to
establish the PPRC paths. You then click Next. In this example we want to establish
PPRC paths from LSS 14. See Figure 34-6.
7. Now you have to select the target LSS. First select in the Storage Complex pull-down
menu the Storage Complex for the ESS. Then select from the Storage Unit pull-down
menu the appropriate Storage Unit and the LSS from the Storage Unit. When you are
finished, click Next to continue. In Figure 34-7 the target LSS on the ESS is LSS 10.
9. In the next panel (see Figure 34-9) you have to select for each source I/O port from the
DS8000 a target I/O port on the ESS. When done click Next.
10.On the next two panels you will be asked whether you want built a Consistency Group and
to verify the information you entered during the process. After you verify the information
you entered, click Finish to establish the RMC paths.
522 IBM System Storage DS8000 Series: Copy Services in Open Environments
11.Figure 34-10 shows the two RMC paths we have established for LSS 14. To manage the
established paths (for example, delete path) you can select the path you want to manage
and then select from the pull-down menu the appropriate action you want to perform.
34.3.3 Establish logical paths between DS8000 and ESS 800 using DS CLI
You can also use the DS CLI to establish logical paths. One additional step is that you need
to know the WWNN of the target storage image. The three commands we will use to establish
a path are lssi, lsavailpprcport, and mkpprcpath.
Determining the WWNN of the ESS 800 using the ESS Specialist
To do this:
1. Point your browser at the IP address of either cluster of the ESS 800.
2. On the opening screen, click the ESS Specialist button.
3. You will receive a logon prompt. Log on to the ESS Specialist using a ESS Specialist user
ID and password.
4. On the welcome screen, you will see the WWNN, as shown in Figure 34-11. You need to
write it down.
Figure 34-11 Using ESS 800 Specialist GUI to display the ESS 800 WWNN
Determining the WWNN of the ESS 800 using the ESS CLI
If you have the ESS CLI installed on a PC then you can use the list server command to
display the ESS 800 WWNN.
Note: This is not the DS CLI. ESS CLI is a separate software package that you can get
from your IBM Service Representative if you do not already have it.
An example of the command syntax is shown in Example 34-6. This technique has the
advantage that you can copy and paste the output.
Example 34-6 Using ESS CLI to display the ESS 800 WWNN
C:\Program Files\ibm\ESScli>esscli -u storwatch -p specialist -s 10.0.0.1 list server
524 IBM System Storage DS8000 Series: Copy Services in Open Environments
Server Model Mfg WWN CodeEC Cache NVS Racks
---------- ----- --- ---------------- -------- ----- ----- -----
2105.22399 800 013 5005076300C09517 2.4.3.79 32GB 2GB 1
Note: When issuing commands that refer to an ESS 800, the ESS 800 serial number is
only five digits, not seven as the DS6000 or DS8000 are. So in our examples, the serial
number syntax we use is IBM.2105-22399, not IBM.2105-1322399.
In Example 34-8, we first establish a single path between LSS 12 on the DS8000 and LSS 15
on the ESS 800. We then display the paths available for LSS 12 on the DS8000.
In Example 34-9 we add an additional path between LSS 12 on the DS8000 and LSS 15 on
the ESS 800. Note that we include the existing path when creating a new path. Otherwise, the
existing path is removed and only the new path will be available for use.
To establish paths where the ESS 800 is the source, we should connect to the ESS 800 using
the DS CLI and follow the same process, but specify the DS8000 as the remote device.
Important: When you connect using the DS CLI to the DS8000 HMC, the DS8000 is the
local device and the ESS 800 is the remove device. If you connect using the DS CLI to the
ESS 800, then ESS 800 is now the local device and the DS8000 is the remote device.
Attention: The rmpprcpath command will remove all paths between the source and target
LSSs. To just reduce the path count, use the mkpprcpath command specifying only the
paths you wish to continue using.
34.4.1 Managing Metro Mirror or Global Copy pairs using the DS GUI
In this example we show how you can establish Metro Mirror volume pairs between a DS8000
and an ESS 800 with the DS Storage Manager. In this example we create two Metro Mirror
pairs. Volumes 1401 and 1402 from the DS8000 are the source volumes, and the target
volumes are 1000 and 1001 from the ESS. We would follow the same method to set up a
Global Copy pair, except that Global Copy would be selected in Figure 34-17 on page 529.
1. Click Real-time manager.
2. Click Copy services.
3. Click Metro Mirror.
4. Select the Storage Complex, Storage Unit, and Storage Image on which the Metro Mirror
source volumes are.
526 IBM System Storage DS8000 Series: Copy Services in Open Environments
5. Click Create; see Figure 34-12.
The next panels guide you through the process of creating Metro Mirror volume pairs. In
the last panel, you have the opportunity to review the information you entered before you
establish the Metro Mirror pairs. Also, during the process, you can, at any time, go back to
modify specifications that you have already done, or you can cancel the process.
6. You have the option to choose between Automated volume assignment and Manual
volume pair assignment; see Figure 34-13. If you click the Automated volume pair
assignment, the first selected source volume is paired automatically with the first selected
target volume. If you click Manual volume pair assignment, you must select each
specific target volume for each selected source volume. In this example we assign the
volume pairs manually.
8. In the next panel, select the storage complex and storage unit for the LSS with the target
volumes. Then select the target volume for the first source volume. When you are
finished, click Next, as shown in Figure 34-15. To expand your choices, you must select
the small blue boxes.
528 IBM System Storage DS8000 Series: Copy Services in Open Environments
9. Next select the target volume for the second source volume, as shown in Figure 34-16. To
expand your choices, you must select the small blue boxes.
10.In the next panel you can specify various copy options, as shown in Figure 34-17. In this
example we selected Metro Mirror under Define relationship type and Perform initial
copy. If you wish to instead use Global Copy, this is where you select it.
In Example 34-10 we create two pairs using the mkpprc command. We then list them using
the lspprc command. We then remove one pair using the rmpprc command.
In Example 34-10 we connected to the DS8000 HMC using the DS CLI, so the ESS 800 is the
remote device. If we wish to establish pairs where the ESS 800 is the source device, we need
to connect to the ESS 800 using the DS CLI. This will make the DS8000 the remote device.
530 IBM System Storage DS8000 Series: Copy Services in Open Environments
Important: If you specify the wrong -remotedev or you are using a profile where a different
remote device from the one you intended to work on is specified, you may get an error
message, CMUN03057, saying you are specifying an invalid subsystem ID. This may be
because you are specifying the wrong remote device serial number. If you have multiple
potential remote devices, do not specify a remotedev in your DS CLI profile.
In Example 34-10 on page 530 we create two pairs using the mkpprc command. We then list
them using the lspprc command. We then pause one pair using the pausepprc command.
532 IBM System Storage DS8000 Series: Copy Services in Open Environments
ESS 800 using the DS CLI is also no different; see 8.3, “Local FlashCopy using the DS CLI”
on page 66. In a situation where the ESS 800 is the target device in a PPRC relationship, it is
also possible to create remote FlashCopies on the ESS 800 using the mkremoteflash
command. In each case, creating, changing, and removing the FlashCopy is done using the
same commands.
8. Choose between A single source with a single target and A single source with
multiple targets and click Next.
534 IBM System Storage DS8000 Series: Copy Services in Open Environments
34.6.3 Creating a remote FlashCopy on an ESS 800 using DS CLI
It is also possible to use DS CLI to create a remote FlashCopy on an ESS 800 where the
source PPRC device is a DS8000. In Example 34-14, we connect using the DS CLI to a
DS8000. We determine there are established paths from LSS 14 on the DS8000 to LSS 10
on the ESS 800. We create a Metro Mirror pair from volume 1401 on the DS8000 to volume
1001 on the ESS 800. We then create a remote FlashCopy on the ESS 800 between ESS
800 volumes 1001 and 1002. We then remove the remote FlashCopy.
Example 34-14 Creating a remote FlashCopy where the ESS 800 is the remote target
dscli> lspprcpath 14
Date/Time: 2 November 2005 20:23:23 IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7503461
Src Tgt State SS Port Attached Port Tgt WWNN
=========================================================
14 10 Success FF16 I0000 I000C 5005076300C09517
14 10 Success FF16 I0130 I000C 5005076300C09517
dscli> mkpprc -remotedev IBM.2105-22399 -type mmir 1401:1001
Date/Time: 2 November 2005 20:14:14 IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7503461
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 1401:1001 successfully created.
dscli> mkremoteflash -dev IBM.2105-22399 -conduit IBM.2107-7503461/14 -persist -nocp 1001:1002
Date/Time: 2 November 2005 20:19:47 IBM DSCLI Version: 5.1.0.204 DS: IBM.2105-22399
CMUC00173I mkremoteflash: Remote FlashCopy volume pair 1001:1002 successfully created. Use the
lsremoteflash command to determine copy completion.
dscli> lsremoteflash -dev IBM.2105-22399 -conduit IBM.2107-7503461/14 1001:1002
Date/Time: 2 November 2005 20:19:59 IBM DSCLI Version: 5.1.0.204 DS: IBM.2105-22399
ID SrcLSS SequenceNum ActiveCopy Recording Persistent Revertible SourceWrite TargetWrite
===========================================================================================================
1001:1002 10 0 Disabled Disabled Enabled Disabled Enabled Enabled
dscli> rmremoteflash -dev IBM.2105-22399 -conduit IBM.2107-7503461/14 1001:1002
Date/Time: 2 November 2005 20:20:10 IBM DSCLI Version: 5.1.0.204 DS: IBM.2105-22399
CMUC00179I rmremoteflash: Are you sure you want to remove the remote FlashCopy pair {0}? [y/n]:y
CMUC00180I rmremoteflash: Removal of the remote FlashCopy volume pair 1001:1002 has been initiated
successfully. Use the lsremoteflash command to determine when the relationship is deleted.
You could also reverse the scenario and use the ESS 800 as the source machine in a PPRC
pair and then create the remote FlashCopy on the DS8000.
Note: Commands that work with remote FlashCopies use the -dev parameter to define the
machine on which the FlashCopy is to be performed. Other commands, such as mkpprc
commands, refer to this device as the remote device, using -remotedev. However,
because a FlashCopy must be sent to the remote site and then performed locally there, the
use of the -dev parameter to refer to the remote machine is correct.
Part 9 Solutions
In this part we discuss solutions offered by IBM to assist you in the management, automation,
and control of your Copy Services implementation on the DS8000. We provide an overview of
the solutions, discuss the options available, and give some examples of using the solutions.
In addition to the DS6000 and DS8000, the IBM TotalStorage Productivity Center for
Replication or TPC for Replication can also help manage copy services for the SAN Volume
Controller (SVC), as well as the ESS 800 when using FCP links for PPRC paths between
ESS 800s or between ESS 800 and DS6000 and DS8000. This also applies to FlashCopy.
For installation and configuration details refer to Powering SOA with IBM Data Servers,
SG24-7259.
You must refer to the following manuals when implementing and managing TPC for
Replication configurations:
IBM TotalStorage Productivity Center for Replication User’s Guide, SC32-0103
IBM TotalStorage Productivity Center for Replication Installation and Configuration Guide,
SC32-0102
IBM TotalStorage Productivity Center for Replication Command-Line Interface User’s
Guide, SC32-0104
Note: The IBM TotalStorage Productivity Center for Replication is completely independent
from TPC for Disk, Data, and Fabric.
Design and development emphasis for TPC is on scalability and standards. The approach
based on open standards allows TPC to manage any equipment or solution implementation
that follow the same open standards.
Figure 35-1 shows a split set of products with three components based on former Tivoli
products and a package with two components designed and developed to facilitate storage
replication solutions.
Replication
TPC for Disk TPC for Data TPC for Fabric Replication
Two Site BC
BC Business Continuity
Figure 35-1 TPC software products suite
All three Tivoli products provide a single source and single management tool to cover the
tasks of the storage manager or network manager in their daily business.
TPC for Disk is software based on open standard interfaces to query, gather, and collect
all available data necessary for performance management.
TPC for Data focuses on data management and addresses aspects related to information
life-cycle management.
TPC for Fabric is a management tool to monitor and manage a SAN fabric.
TPC for Replication, as a member of the TPC product family, is a standalone package and
does not build on anything related to the TPC Standard Edition.
In the following paragraphs we cover the basic structures and features of TPC for Replication
and TPC for Replication Two Site BC. This is to explain and discuss Copy Services topics
with the DS6000 and DS8000.
540 IBM System Storage DS8000 Series: Copy Services in Open Environments
35.2 Where we are coming from
Since the advent of copy services functions with IBM storage servers a framework is required
to handle and manage disk storage environments and the various combinations of software,
firmware, and hardware used for replication.
To ensure and guarantee data consistency at the remote or backup site it is even more
important to provide a framework around copy services functions that helps achieve that data
consistency and ease of management.
Other aspects are automation and scalability. Early solution attempts turned out to have
weaknesses that would often surface during actual implementation in real production
environments. Tools offered on an as is basis are not suitable for managing production
environments. Other solutions are perfectly suited for certain host server platforms but cannot
manage other server platforms.
This led to the design and implementation of a framework that is platform independent and
provides the required production attributes such as scalability, stability, and security, and
provides as a standard product also offering the guarantee to be serviced and maintained.
Furthermore, the concept has the potential for enhancements and can evolve over time and
according to user requirements.
TPC for Replication builds on all previous experiences with storage management tools and
framework proposals to organize all aspects of copy services for disaster recovery solutions.
TPC for Replication also address the need for other solutions, which involve copy services
functions available with the DS6000 and DS8000 like data or volume migration, data center
movements, or other projects that require copying or moving data between similar devices
such as the DS6000, DS8000, and ESS 800.
Figure 35-2 is a screen capture from the TPC for Replication GUI showing the different
session types, or Copy Services functions, supported.
Figure 35-2 Management of Copy Services Functions supported by TPC for Replication
Note that TPC for Replication also manages FlashCopy and Metro Mirror for IBM SAN
Volume Controller (SVC).
35.4.1 FlashCopy
FlashCopy is a point-in-time copy of a source volume mapped to a target volume. It offers
various options including the choice between a COPY option, for a background copy through
physical I/O operations in the storage server back-end, or a NOCOPY option.
FlashCopy is available on the DS6000 and DS8000 as well as on the ESS 800. FlashCopy is
also available on the DS4000 family and with the SAN Volume Controller. Note, however, that
the DS400 and SVC use different implementations of the FlashCopy function that are not
compatible with the DS6000. DS8000, and ESS 800 FlashCopy.
542 IBM System Storage DS8000 Series: Copy Services in Open Environments
35.4.2 Metro Mirror
Metro Mirror (MM) is a synchronous data replication mechanism and was previously called
Peer-to-Peer Remote copy (PPRC). I/O completion is signaled when the data is secured on
both the local storage server and the remote storage server. Metro Mirror is popular for two
site disaster recovery solutions.
Metro mirror is available in any combination between DS6000, DS8000, and ESS800.
Synchronous data replication is also available on the DS4000 family and with the SVC, but
again not directly compatible with Metro Mirror on the DS6000, DS8000, and ESS 800.
Global Copy is possible in any combination between DS6000, DS8000, and ESS 800.
Restriction: TPC for Replication does not provide an interface to manage a Global Copy
environment.
GM is also the base for a two-site disaster recovery solution involving any DS6000, DS8000,
or ESS 800. The function as such is also possible between products of the DS4000 family
and the SVC.
MGM is a combination of MM and GM. This is a synchronous data replication between two
sites in a metropolitan area and an asynchronous copy of the same data at a different remote
site, which can be located at any distance.
MGM configurations and operations are currently not supported by TPC for Replication.
A A 1. Failover command
A
Primary
Primary A
Primary
Primary
Primary Primary
P
Primary
P
Primary
2. Failback command
PRIMARY SUSPENDED
A Resynchronisation
A
A
Primary
Primary A
Primary
Primary
Primary Primary
SECONDARY PENDING S
Primary
PRIMARY PENDING P
Primary
OR 2. Failback command
A A
Resynchronisation
A
Primary
Primary A
Primary
Primary
Primary Primary
PRIMARY PENDING P
Primary
SECONDARY PENDING S
Primary
2. Once the outage is over and you decide to resume replicating data between, you have a
choice as whether to re-synchronize from the remote site to the local site or vice versa.
This depends on what is required after updating one of the two sites or even updating both
sites after the suspend action.
544 IBM System Storage DS8000 Series: Copy Services in Open Environments
a. When you decide to re-synchronize from the remote site to the local site the failback
command is pointing to the remote volumes. You need a PPRC path from the remote
site to the local site before you can perform the corresponding commands. Doing this
will resynchronize both sites with the level of data as it is at the remote site. Only the
data that changed at the remote site, from the moment the replication was suspended
and until the failover happened, is replicated from remote to local volumes.
Figure 35-3 on page 544 assumes a Metro Mirror (MM) environment that puts the
corresponding MM pair in PENDING state during re-synchronization, and once
everything is replicated the MM pair goes into DUPLEX state.
b. If you want to re-synchronize both sites with the data level at the local site, issue the
failback command towards the volumes at the local site. This replicates the changed
data since the suspend from the local site to the remote site. Also, this resets the
tracks that were potentially changed at the remote site after the failover.
Figure 35-3 on page 544 assumes a MM environment that places the MM pair in
PENDING state during the re-synchronization phase. Once all changed data is
replicated, the MM pair goes into DUPLEX state.
The option to re-synchronize from either site is possible due to the availability of change
bitmaps maintained by the DS6000 or DS8000 at both sites.
A
P3
Primary
Primary
PPRC path S3 Copy Set
Primary Secondary
PPR link
A
P2
Primary
Primary
PPRC path S2 Copy Set
Primary Secondary
PPR link
A
Primary
A
Primary
Primary
P1
Primary PPRC path S1
Primary Copy Set
Primary Secondary
546 IBM System Storage DS8000 Series: Copy Services in Open Environments
Figure 35-5 represents two copy sets. In this illustration, each copy set contains three
volumes, which indicates a Global Mirror configuration. Note that the third volume is a kind of
a journal volume and is the FlashCopy target volume covering for data consistency in a
Global Mirror setup.
Global Copy
FlashCopy
A A
Primary Primary
P2
Primary S2
Primary
Primary Secondary
Copy Set
Primary
Primary
J2
Tertiary
PPRC links
Global Copy
A A FlashCopy
Primary
Primary
P1
Primary S1
Primary
Primary Secondary
Primary Copy Set
Primary
J1
Tertiary
A session is a logical concept that gathers multiple copy sets, representing a group of
volumes with the requirement to provide consistent data within the scope of all involved
volumes. Commands and processes performing against a session apply these actions to all
copy sets within the session. Again this is similar to a Global Mirror for System z (XRC)
session.
Session A
P3
Primary
Primary PPRC path S3 Copy Set
Primary Secondary
1
A
P2 PPRC path Copy Set
Primary
Primary S2
Prim ary Secondary
A A
Primary Copy Set
Primary
P1
Primary PPRC path Primary
S1
Primary
Primary Secondary
Session A
Primary
PB
Primary PPRC path
A
Primary
Primary
SB
Primary
Copy Set
Primary
2
Secondary
A A
Primary Copy Set
Primary
PA
Primary PPRC path Primary
SA
Primary
Primary Secondary
Metro Mirror primary volume P1 in the second storage server, and volumes P2 and P3 in the
first storage server with their corresponding Metro Mirror secondary volumes are grouped
together in a session, session 1. Such a session can contain copy sets that span across
storage server boundaries.
Metro Mirror primary volumes PA and PB with their counterparts at the remote site belong to
a different session, session 2.
Note that all application-dependent copy sets must belong to the same session to guarantee
a successful management and to provide consistent data across all involved volumes within a
session. In this context we recommend having application-dependent volume granularity
within the scope of an LSS. Or, in other words, volumes or copy sets that require consistent
data and can be subject to a freeze trigger ought to be grouped in one or more LSSs and that
LSS must not contain other copy sets. This because the scope of the freeze function is at the
LSS level and affects all volumes within that LSS.
548 IBM System Storage DS8000 Series: Copy Services in Open Environments
35.6 TPC for Replication session types
There are two TPC for Replication editions, which include different levels of Copy Services
functions.
FlashCopy
FlashCopy includes all functional flavors of FlashCopy. This includes FlashCopy with:
Background copy or no background copy.
Convert no background copy to background copy.
Add persistent to a FlashCopy relationship.
Establish incremental FlashCopy to apply only changed data since a previous FlashCopy
operation.
Metro Mirror
Metro Mirror is between a local site (site 1) and a remote site (site 2) in a uni-directional
fashion. It also allows switching the replication direction from site 2 to site 1. This may happen
through failover/failback operations.
Global Mirror
Global Mirror implies three volumes for each Global Mirror copy set. Again, this happens
between a local site (site 1) and a distant site (site 2) in a uni-directional way. It is possible to
reverse the replication direction but still only in a uni-directional fashion.
Metro Mirror
Metro Mirror may be established from site 1 to site 2 or vice versa. It allows also
failover/failback operations for any copy set at any time.
Global Mirror
The Advanced Edition allows the Global Mirror to operate in either direction at the same time
(bi-directional). It may also swap sites with failover/failback operations.
Global Mirror builds on three volumes per copy set. TPC for Replication allows also managing
a configuration that only replicates data through Global Copy in the opposite direction than
Global Mirror did before.
Important: Do not manage through another software, Copy Service volume pairs which
are already managed through TPC for Replication.
A copy set contains the volumes that are part of a Copy Services relationship. Metro Mirror
knows two volumes per copy set. Global Mirror requires three volumes per copy set.
FlashCopy again consists of two volumes per copy set.
550 IBM System Storage DS8000 Series: Copy Services in Open Environments
Terminology is slightly different from what you may be used to. For example, Metro Mirror
uses a primary or source volume at the sending site and a secondary or target volume at the
receiving end. Such a pair is now called a copy set.
FlashCopy usually used to mention source and target volume to identify a FlashCopy pair, in
contrast to RMC, which designates volumes in that pair as primary and secondary volumes.
Again such a FlashCopy volume pair is now a copy set.
Global Mirror involves three volumes per copy set. Global Mirror relies on Global Copy and
FlashCopy. Therefore we have a Global Copy primary volume and a Global Copy secondary
volume. The Global Copy secondary volume has another role as FlashCopy source volume.
The third volume is then the FlashCopy target volume. All three volumes are part of a Global
Mirror copy set.
When a host volume becomes the target volume of a Copy Services function it is usually no
longer accessible from the application host. FlashCopy target volumes can be considered as
an exception.
A A A
APrimary
Primary APrimary
Primary APrimary
Primary
A Primary
Primary A Primary
Primary A Primary
Copy Set
Primary A
Primary
Primary
Primary
Primary Session 1 Primary
Primary APrimary
Primary
APrimary
Primary
Primary
Primary
A A
A Primary
Primary A
A Primary
Primary Session 2 Primary A Primary
Primary
Primary Primary Primary
Primary Primary
A A
Primary Primary
Primary Primary
A A
A Primary
Primary Session 3 A Primary
Primary
Primary Primary
APrimary APrimary
Primary Primary
Primary Primary
A A A
Primary
Primary Session 4 Primary
Primary
Primary
Primary
There is also no limit to the number of sessions that TPC for Replication can manage.
Figure 35-7 shows four sessions managing different copy sets. The first session, session 1,
implies a Metro/Global Mirror configuration that is not supported (at the time of writing this
book) by TPC for Replication.
The second session, session 2, is a Global Mirror session between an intermediate site and a
remote site with two copy sets.
The third session, session 3, is a Metro Mirror configuration between the local and
intermediate site.
The session at the bottom in Figure 35-7, session 4, is a Global Mirror session between the
local and remote sites.
552 IBM System Storage DS8000 Series: Copy Services in Open Environments
Figure 35-8 shows how the TPC for Replication server connects to the storage server that is
going to be managed by TPC for Replication servers. It does not show the connectivity of the
TPC for Replication server to the network through which the user connects with the browsers
to the server itself.
Primary
Primary DB2 Websphere
Windows
Database Replication Manager Server Linux
AIX
IP Communication
Services
IP network
Ethernet ports
Series p Series p PowerPC PowerPC
FCP ports
A A
APrimary
Primary
APrimary
Primary
A Primary
Copy Set
Primary A A
Primary
Primary A
Primary
Primary
Copy Set
Primary A
Primary
A
Primary
Primary
Primary A
Primary
Primary
Primary
Primary
APrimary
Primary
Primary Primary
Primary
PPR FCP links
A A
A Primary A A
Primary A Primary A Primary
Primary
Primary
Primary
Primary
Primary Primary A Primary
Primary
Primary Primary Primary
Primary
DS8000 DS6000
Figure 35-8 TPC for Replication system overview
TPC for Replication server contains, besides the actual functional code, a DB2 UDB and
WebSphere® software.
IP network
Ethernet ports
PowerPC PowerPC
server0 server1
server0 server1
DS8000
The actual connectivity between the TPC for Replication server and the storage servers is
based on Ethernet networks and connects to particular Ethernet ports in the System p™ in
the DS8000. This particular Ethernet card is a new card and slides into the first slot out of
these four slots in the p570. Note that Figure 35-9 shows the DS8000 rear view because
these slots are only accessible from the rear site of the DS8000.
These features are chargeable and carry a minimum monthly maintenance charge.
To configure the Ethernet ports, Release 2 microcode is required for the concerned DS8000.
This is a code bundle that starts with 6.2.xxx.xx.
554 IBM System Storage DS8000 Series: Copy Services in Open Environments
Port numbers on the first card are I9801 and I9802. This is the card that installs in server0.
Port numbers on the second card are I9B01 and I9B02. This is the card that installs in
server1. Note that only the first port on each card is currently used.
Communication through these ports uses static IP addresses. DHCP is not supported.
You may configure these new ports either through the GUI or the DSCLI.
The GUI provides a new panel to configure the required IP addresses. This panel is in the
GUI path of the storage image. Select a storage image, then select Configure network
ports from the Select Action pull-down (note that this option is only presented when the
network card for TPC for Replication is actually installed). This is shown in Figure 35-10. The
Configure Network ports panel will then display.
The DSCLI provides the following commands to manage these new network ports:
lsnetworkport -l
This command shows the server associations, physical port location, and all IP address
settings for all ports on the queried Storage Image Facility. See Example 35-2 on
page 556.
shownetworkport
This shows the server association, physical location, and IP addresses for a particular port
on the Ethernet card. See Example 35-3 on page 556.
Example 35-2 shows an output example of lsnetworkport -l, which provides an overview of
all available Ethernet ports on the concerned storage image facility.
556 IBM System Storage DS8000 Series: Copy Services in Open Environments
Storage Management console connects to because the DS6000 controller or server card
contains only a single Ethernet port.
For the DS6000 there is only a single Ethernet port per cluster or server card. This port is
shared between the Replication Manager server or servers as well as with one or two Storage
Management consoles.
Note that the internal and potential external HMCs connect to the DS8000 through different
ports than the Replication Manager servers.
The communication between the Replication Manager Server and the DS infrastructure is
direct, as shown in Figure 35-9 on page 554. It is interesting to note that this is different from
when the Replication Manager communicates with the SAN Volume Controller (SVC).
Between the Replication Manager Server and the SVC nodes is a SVC CIMON based
console, which is part of the standard SVC master console.
For more details refer to Powering SOA with IBM Data Servers, SG24-7259.
FREEZE 3
2
LSS LSS
LSS
1 LSS
LSS LSS
LSS LSS
LSS LSS
LSS LSS
Primaries
Session Secondaries
TPC for Replication server listens to incidents from the storage servers and takes action
when being notified of a replication error from the concerned storage server.
Figure 35-11 implies a replication error in an LSS that belongs to the session. The TPC
server receives a corresponding SNMP trap message from the concerned storage server.
TPC for Replication server then issues a freeze command to all LSSs that are part of the
concerned session. This implies a suspend of all PPRC pairs or copy sets that belong to this
session. During this freeze process, write I/Os are held until the freeze process ends and the
TPC server communicates to the storage server to continue processing write I/O requests to
the concerned primary volumes in the session.
After that, write I/O can continue to the suspended primary volumes. However, both sites are
not in sync any longer but the data on the secondary site is consistent (power drop
consistent). This is a freeze-and-go policy.
558 IBM System Storage DS8000 Series: Copy Services in Open Environments
35.12 TPC for Replication heartbeat
Because the connectivity between the TPC for Replication server and the storage servers
that the TPC server is managing can fail, the firmware in the storage server wait for a
heartbeat signal from the TPC server. TPC for Replication can enable this heartbeat in the
corresponding LSS for Metro Mirror sessions.
1 FCP ports
2 FREEZE
LSS LSS
LSS LSS
Primaries
Session Secondaries
Figure 35-12 LSS heartbeat triggers freeze when connectivity to server fails
Figure 35-12 implies a failing connectivity between the RM server and a primary storage
server. Once this heartbeat is set and the storage subsystem cannot communicate its
heartbeat information to the RM sever, the storage server internally triggers a freeze to its
involved LSS and their primary volumes. Because this heartbeat is a timer scheduled
function, it is based on the Consistency Group time-out value of each LSS that contains
volumes that belong to the concerned session. This session-based heartbeat timer expires
after the lowest time-out value of all concerned LSSs.
With more than a storage server involved, the RM server issues freeze commands to all other
LSS pairs in the affected session once the heartbeat expires and the RM server could not
receive heartbeat information from any one of the involved storage servers.
The disconnected storage server or LSS resumes I/O after their Consistency Group time-out
has expired. LSS or storage servers that may still be connected to the RM server receive a
corresponding freeze command from the RM server followed by a run command when a
freeze-and-go policy has been selected.
Note that Extended Long Busy (ELB), automation window, and freeze period are synonyms
for Consistency Group time-out in the above context.
SUSE Linux Enterprise Server 9 SP2 (Note that SUSE Linux does not support the Two
Site BC configuration.)
Red Hat Enterprise Linux RHEL4 AS 2.1 (For Replication Two Site BC: Red Hat
Enterprise Linux RHEL4 Update 1 SLES9 SP2)
AIX 5.3 ML3
Note that for a TPC for Replication Two Site BC configuration that involves two servers, it is
possible to run TPC for Replication under two different operating system platforms.
When TPC for Replication runs on AIX the following minimum hardware configuration is
suggested:
Server p, IBM POWER4™ or IBM POWER5™ processor, 1 GHz
2 GB RAM memory
10 GB free disk space
560 IBM System Storage DS8000 Series: Copy Services in Open Environments
Disk space is required to hold for data in DB2 databases and WebSphere Express
Application Server code besides the actual TPC for Replication server code.
You may refer to the following Web site for the latest information about hardware
requirements:
http://www-03.ibm.com/servers/storage/support/software/tpcrep/installing.html
TPC for Replication provides a graphical user interface to manage and to monitor any Copy
Services configuration and Copy Services operations. This GUI is Web browser based and
does not rely on any other product like TotalStorage Productivity Center and IBM Director.
Figure 35-14 GUI URL determined during Replication Manager install time
The GUI is simply called through an Internet browser like MS Internet Explorer® or Mozilla
Firefox, is intuitive, easy-to-use and performs very well. The panel structure is not too deep
and allows you to quickly transition to any application through a hyperlink-based menu.
Login
LAN
Hardware Lavers
IP network
DS8000 DS6000
The Web-based client that runs on the user’s machine contains the GUI client, a presentation
component. This software piece on the user’s machine is highly optimized to minimize data
traffic over the LAN to the RM server.
The GUI server component runs on the RM server machine that contains the interface to the
function code in the connected storage servers and is usually closely installed to the storage
servers that interface with the RM server.
562 IBM System Storage DS8000 Series: Copy Services in Open Environments
35.15.1 Connect to the TPC for Replication GUI
Connect to the GUI by specifying the IP address of the TPC for Replication server in your
Web browser. This will present you with the sign-on panel, as shown in Figure 35-16. Once
you sign out from the RM server the same panel is also displayed.
user name
password
Specify a user ID as a text string on the UserID filed and a password in a hidden text field.
User IDs are defined and set up in the RM server system.
This health panel displayed an overall summary of the Replication Manager system status. It
shows information very similar to what is also shown in the small box in the left lower corner
of this panel. This small health overview box at the lower left corner is always present.
However, the Health Overview panel provides more details. It provides a first overall status of:
Sessions
Connected storage subsystems
Management servers
Figure 35-17 reports that all sessions are in normal status and working fine. There is no high
availability server environment, and one or more storage servers cannot be reached by the
Replication Manager, as they have been defined before.
The upper left box in this panel labeled My Work provides a list of applications that you can
use to manage various aspects of a copy services environment:
Health Overview
This is the currently displayed panel, as Figure 35-17 shows.
Sessions
This hyperlink brings you to the application that manages all sessions. This is the
application that you will use the most.
Storage Subsystems
Here you start when you define storage servers to the RM server that are going to be used
for Copy Services.
564 IBM System Storage DS8000 Series: Copy Services in Open Environments
ESS/DS Paths
This link allows you to manage everything that is related to PPRC path management.
Management Servers
This links lead you to the application that manages the RM server configuration.
Advanced Tools
Here you may trigger to collect diagnostic information or set the refresh cycle for the
displayed data.
Console
This link opens a log that contains all activities that the user did and their results.
Each session comprises a number of copy sets that may be distributed across LSSs and
physical storage servers. The session name functions as a token that is used to apply any
action again the session or all the volumes that belong to that session.
Figure 35-20 displays a next possible step to perform an action. After selecting the session
GM_Session1, as shown in Figure 35-18 on page 565, you may select any action from the
action list shown in Figure 35-20.
This may be an action against the entire session, like suspending all volumes within the
session. It is also used to modify an existing session and add or remove copy sets.
566 IBM System Storage DS8000 Series: Copy Services in Open Environments
35.15.4 Storage Subsystems panel
The following panel (Figure 35-21) displays all storage subsystems currently connected to the
RM server.
Panel heading
Hyper links
RM summary
Using the Add Subsystem button you can define another storage subsystem to the RM
server.
The panel used to add a new server is shown in Figure 35-23 on page 568.
Figure 35-22 Select storage subsystem and select View/Modify Details action
The selected storage subsystem connectivity details are now displayed, as shown in
Figure 35-23.
568 IBM System Storage DS8000 Series: Copy Services in Open Environments
You usually use the Storage Subsystems application only to connect a storage server to the
RM server. Figure 35-23 on page 568 also shows the standard port used port used by the RM
server to communicate with the storage server. All the other fields are self explanatory.
Clicking the Manage Path button will trigger the path wizard to help you define a new PPRC
path or remove existing PPRC paths.
You may select any path here and the only available action in this case is to then remove the
selected paths.
570 IBM System Storage DS8000 Series: Copy Services in Open Environments
Figure 35-26 shows only one server named Stops, which is an active server.
When it exists, a second row will show the potential second server, which is in the status
standby. A panel with two servers has a slightly different appearance than what is shown in
Figure 35-26.
Use this panel for basic operations such as defining a server as standby or to take over in the
event of a disaster.
In the case of two servers, each server manages its own DB2 database. The communication
between both servers is performed through the LAN to which both servers are connected.
Specific tasks in this context are tasks to create a diagnostic package or to change the
automatic refresh rate of the GUI. A third task is to enable or to disable the heartbeat, which
happens between the Replication Manager server and the connected storage servers. Aee
35.12, “TPC for Replication heartbeat” on page 559.
The diagnostic package contains all RM logs and its location on the RM server is shown on
the screen.
The browser refresh rate is currently a value between 5 and 3600 seconds. The default value
is 30 seconds.
Besides the commands this log shows also whether the command succeeded and a
message number functions at the same time as a hyper link to more detailed text about the
result of the concerned command execution.
As with the DSCLI for DS8000 and DSCLI for DS6000, the CSMCLI command structure is
similar between all three CLI products like mk... for make, ch... for change, and.
CSMCLI is also invoked in the same fashion as the DSCLI for the DS products and provides
three different modes:
Singe-shot mode
Interactive mode
Script mode
572 IBM System Storage DS8000 Series: Copy Services in Open Environments
To execute more than one command you may start the CLI program and perform the
commands at the shell prompt, as Example 35-5 shows.
The third mode is the script mode to run commands out of a file.
In contrast to DSCLI for the DS storage servers, the CSMCLI does currently not use a -profile
option.
eRCMF is a standalone tool that provides a management layer for ESS800, DS8000, and
DS6000 Copy Services in 2-site or 3-site topologies. It is a common management tool for
both z/OS and open system environments. The Copy Services functions managed by
eRCMF are Metro Mirror, Global Copy, and Global Mirror. At each site, a FlashCopy of the
local volumes can also be managed.
This solution simplifies the process of repairing inconsistent Remote Copy pairs after an error
with the copy process. eRCMF assists the user in automating the Remote Copy and
FlashCopy processes.
Attention: Because of its complex implementation, IBM Global Services will work with
your IT department to install and configure the solution to match your unique business
environment. The storage servers, servers, and WebSphere software have to be ordered
through the normal process. IBM Global Services provides eRCMF software and experts
that support the implementation, education, and skill transfer for this solution.
The solution provides basic commands through the FO/FB Session Manager Proxy, which
utilizes the ESS Open API functions to accomplish either a planned or unplanned FO/FB
sequence. eRCMF commands can be incorporated in any user scripts or utilities that may
further link users’ overall data center operations. However, it is the user’s responsibility for
creation, maintenance, and any liabilities associated with such scripts or utilities. Therefore,
eRCMF offers a simple way to implement Remote Copy disaster recovery procedures via
command line. eRCMF applies the commands to a set of configuration files that describe the
disk subsystem configuration (paths and sets of volumes).
The solution does not depend on the operating system, as no client installation is required.
36.2 Overview
In a Business Continuity solution based on disk Remote Mirroring, we want to restart a
database application following an outage without having to restore the database. This
process has to be consistent, repeatable, and fast, measurable in minutes.
A database recovery where you have to restore the last set of Image Copy tapes and apply
the log changes to bring the database up to the point of failure is a process that can be
measured in hours or even days. And this is what you want to avoid in a Tier 4 or later
576 IBM System Storage DS8000 Series: Copy Services in Open Environments
solution. However, in a real disaster (fire, explosion, earthquake) you can never expect your
complex to fail all at the same moment. Failures will be intermittent and gradual, and the
disaster will occur over many seconds or even minutes. This is known as the Rolling
Disaster. The architecture of a viable disk mirroring disaster recovery solution must avoid
data corruption caused during a rolling disaster.
In any operating system, the sequence in which updates are being made is what maintains
the integrity of the data. If that sequence is changed, data corruption will occur. The correct
sequence must be maintained within a volume, across volumes, and across multiple storage
devices. For instance, in Figure 36-1 we show the relationship between a database and its
log, which demonstrates the requirement for maintaining I/O integrity. Data consistency
across the storage enterprise must be maintained to insure data integrity.
Primary
Primary Secondary
Secondary
Database
2. Update Database
Volume
The order of dependent writes across volumes must be maintained at remote locations.
Failure to do so results in data corruption. The data consistency is lost.
Chapter 36. IBM TotalStorage Rapid Data Recovery for UNIX and Windows 577
In Figure 36-2 we illustrate this concept with an example. When a database update is about to
take place, the change is first logged in the database log files at both the primary and
secondary volumes (step 1). Assume the database datafile is updated at the primary volume,
but the update does not reach the remote volume that contains the mirrored data file. The
primary location is not aware of the write failure to the secondary volume (step 2). The
database update is marked complete in the log files at both the primary and remote locations
(step 3). The result is that the secondary site log files say the update was done, but the
updated data is not in the database at the secondary location. There is no way to know that
the data was corrupted.
Primary
Primary Secondary
OK 1. Log Update
B C
OK
L M 3. Mark Log
Complete
X Y
Application
Database
B C
OK 2. Database Update
L M
X Y
So, the issue is that the disk subsystem cannot do it all and we must avoid this Rolling
Disaster system problem.
For this reason it is strongly recommended that, at a minimum, Consistency Groups be put in
place for any mirroring solution. Consistency Groups are an implementation of technology
that assists with the consistency of application data capable of dependent writes. To
guarantee a fully consistent remote copy, multiple volumes require a Consistency Group
functionality. ESS and DS6000/DS8000 Remote Copy microcode already have the
Consistency Group function for both z/OS and open systems. SAN Volume Controller Metro
Mirror’s and DS4000 Remote Mirror’s microcode also have a Consistency Group function for
open systems.
If any volume within a Consistency Group is unable to complete a write to its counterpart in
the Remote Copy relationship, an Extended Long Busy (ELB) for mainframe environments or
a SCSI Queue Full condition for open systems will be issued, preventing further writes to any
of the volumes within the Consistency Group. This wait period is the perfect time to issue a
freeze to all volumes involved to maintain consistency. Figure 36-3 on page 579 illustrates
this mechanism. If a write is unable to complete, the storage subsystem will not back out
incomplete transactions on its own. Instead, the application will need to recognize that the
transaction was incomplete and take the appropriate actions. Once the storage subsystem
578 IBM System Storage DS8000 Series: Copy Services in Open Environments
holds application I/O to the affected primary volumes, the write dependent mechanism of the
application prevents the Remote Copy secondaries from becoming inconsistent.
Consistency Group
control function
SNMP trap
B C
L M
1. Mirroring failure
2. ESS suspends affected primary
X Y
Application
When using freeze through the DS Storage Manager (Graphical User Interface) or ICKDSF,
it will take effect on each LSS individually. This is useful for creating a Point-in-Time Copy of
the data, but because of slight delays between the issuing of each iteration of the freeze
command it is unsuitable for preventing Rolling Disasters and should be done at periods of
low utilization to ensure that the secondary data can be used.
When Remote Copy is used in conjunction with automation, such as the Geographically
Dispersed Parallel Sysplex™ (GDPS) or enterprise Remote Copy Management Facility
(eRCMF) service offerings from IBM Global Services, a freeze command can be
simultaneously issued to all LSSs within the configuration. This ensures globally consistent
data across all LSSs in the secondary copy of data during a disaster.
Chapter 36. IBM TotalStorage Rapid Data Recovery for UNIX and Windows 579
The goal of the TotalStorage Rapid Data Recovery for UNIX and Windows solution, based on
the combination of ESS or DS6000 or DS8000 Copy Services with enterprise Remote Copy
Management Facility (eRCMF), is to protect your data from being a mirror of a dying scenario.
Using a Remote Copy Consistency Group, this solution freezes your environment at a known
point instead of mirroring literally hundreds of time-offset failures in a short amount of time.
Note: We strongly recommend that the RCMF servers be installed and run from different
physical sites to ensure their backup capability.
eRCMF is a multi-site disaster recovery solution that manages data on volumes for Fixed
Block (open systems hosts) and CKD (z/OS). The solution does not depend on the operating
system, as no client installation is required.
t a pe
3 Automation for Servers: z/OS, UNIX, Linux, Windows and
Autonomic
Business value
heterogeneous environments.
Services
eRCMF is built on the technologies of Remote Copy, Java, WebSphere, and ESS Network
Interface. This is because ESS 800, DS6000, and DS8000 share the same Copy Services
functions that you can mix these storage servers in this solution. With DS8000 and DS6000,
no Copy Services Server is required since each storage server can handle the Copy
Services.
36.3 Architecture
eRCMF is a 2-site or 3-site solution. According to the diagrams in Figure 36-5 on page 581
and Figure 36-6 on page 581, one PCM is placed in each data center. The PCM in the
primary data center runs the master process. This process is the interface into eRCMF. It
handles all queries and commands for local processes through either a command-line or
socket interface. It also manages commands and queries from the Backup Process.
A second PCM can be placed in the secondary (or backup) data center or, in a 3-site
strategy, in the tertiary (or remote) data center. This PCM runs the Backup Process, which
580 IBM System Storage DS8000 Series: Copy Services in Open Environments
maintains the state information of the configuration. This information is transferred to the
second PCM in the form of logs from the master process. These logs are then used to update
state information in case the backup process must take control of the configuration, as well as
for documentation purposes if there is an alternate site failure.
Journal Journal
Volumes Volumes
TCP/IP / SNMP
eRCMF eRCMF
eRCMFconfig.properties eRCMFconfig.properties
Enterprise.dat Enterprise.dat
VolumeSet.dat VolumeSet.dat
Hosts.dat Hosts.dat
Authentication.file Authentication.file
eRCMF eRCMF
eRCMFconfig.properties eRCMFconfig.properties
Enterprise.dat Enterprise.dat
VolumeSet.dat VolumeSet.dat
Hosts.dat Hosts.dat
Authentication.file Authentication.file
Chapter 36. IBM TotalStorage Rapid Data Recovery for UNIX and Windows 581
The master process continuously monitors and manages the entire environment at two
levels:
Enterprise level
VolumeSet level
The distinction between these levels is primarily the scope of actions performed at each level.
The Enterprise level represents the entire monitored environment, which consists of two or
three data processing locations containing storage managed by eRCMF. The VolumeSet
level corresponds to a set of storage server volumes, which must be maintained in a
consistent state and managed using the same type of copy operations. Some examples of a
VolumeSet are the volumes used by a set of applications, the volumes allocated to a set of
servers, or the members of a Volume Group.
Management at the Enterprise level affects the entire monitored environment. These actions
typically affect every volume of every VolumeSet being managed by eRCMF. For example,
the freeze command will cause suspension of Remote Copy pairs between the storage
subsystems on multiple sites, as you can see in Figure 36-7.
eRCMF adds a central control function to coordinate the Remote Copy Freeze function and
stops transmission of data to the remote site consistently across single/multiple frames.
When this function is centrally coordinated, data is consistent. Data consistency allows a very
fast, very consistent restart time at the secondary site.
yNotification of mirroring
error
•eRCMF issues freeze command to all Storage
yCU holds I/Os to eRCMF Servers
suspended device until •suspends all PPRC pairs
eRCMF can handle •holds I/O until told to continue
condition
•Based on policy, eRCMF will then allow host
operation to resume or continue to hold them
Control Control
Host Unit
Control Unit
Control
During
Host Unit Unit
Host
Control Control
Unit
Control Unit
Control
Host
Unit Unit
Control Control
Control
Unit Control
Unit After, depending on policy:
Unit
Control Unit
Control
Unit
Control Unit
Control •Both sites are suspend but in sync
Unit Unit
•Sites are not insync, however data
to a site is consistent
582 IBM System Storage DS8000 Series: Copy Services in Open Environments
Global Mirror
– The storage server periodically forms Consistency Groups.
– When recovering to the recovery copy, software such as eRCMF examines the states
of the Global Mirror relationships and directs the storage server through the steps
necessary to recover that data to the last committed consistent copy.
It should be noted that eRCMF neither performs an automated recovery nor an automated
restart of systems in the remote site. Instead, eRCMF's freeze performs a controlled site
split only. If a higher level of automation is required, eRCMF provides interfaces such as
execution scripts or a command-line interface, which can be utilized and customized. When
this occurs, mirroring between site 1 and site 2 stops, ensuring the consistency of data.
Further response by the automation is determined by policies that are selected based on
individual business requirements:
Freeze and Go quickly executes the freeze to ensure data consistency, but then
immediately allows I/Os to continue. This allows processing to continue in the production
data center while maintaining a consistent copy of the data in either the backup or remote
data center. Meanwhile, the business is able to determine the cause of the event and
whether it is a disaster. If a disaster has not occurred, production has not been impacted
by an unnecessary site switch. If a disaster has occurred, then the data since the time of
the site split may need to be recreated if recovering to a backup or a remote data center.
Freeze and Stop quickly executes the freeze to ensure consistency but, rather than
immediately allowing the I/Os to continue, causes the storage server to block all further
I/Os until told otherwise. This ensures that no transactions, other than those in-flight at the
time of the event, need to be recreated following a disaster. If the event turns out not to be
a disaster, however, the production will be halted until the I/Os are freed.
http://www-1.ibm.com/services/us/index.wss/so/its/a1000110
Chapter 36. IBM TotalStorage Rapid Data Recovery for UNIX and Windows 583
584 IBM System Storage DS8000 Series: Copy Services in Open Environments
37
This services offering, available by RPQ through IBM Systems Group, enables high
availability and disaster recovery in the Microsoft environment by incorporating the benefits of
Metro Mirror.
IBM TotalStorage Continuous Availability for Windows is a disaster recovery solution for
Microsoft Cluster server applications using automated cluster failover services combined with
IBM storage subsystems using Metro Mirror for remote site mirroring and automated failover
of data. It can allow for zero or little data loss and little or no downtime, and it provides for
complete site server application and data disaster recovery across two geographically
dispersed sites.
It monitors the health states of the applications, network, and Metro Mirror storage to
determine automatic actions for failover to the secondary site.
Solution highlights
The main solution highlights and benefits are summarized in the following list:
Improve continuity of business.
Improve protection of critical business data.
Reduce business risk.
Provide flexibility for the transfer of operations between sites with minimal disruption.
Improve business resiliency.
Stay competitive, maintaining market readiness by managing risk more efficiently.
37.2 Overview
When you are using Microsoft Clustering Services with Windows 2003 Enterprise Edition
Servers, you may not only want to prevent your business from a disaster on a single node
participating in the cluster but might also want to prevent a disaster on the disk subsystems
shared by the nodes. The solution is to have a Geographically Dispersed Solution where both
servers and storages resources are redundant and installed on different sites. In such a case,
you need to mirror your shared disks between both sites. Since you cannot use any RAID
controller card in a SAN environment, two choices are possible. Either you ask the Windows
system to do it or you ask the disk subsystem to do it.
The first solution is based on a system mirror. When you use a system mirror, the
synchronization process is supported by the resources of your cluster server node. Because
Microsoft Clustering Services does not support dynamic disks, the only way to perform it is to
add a software layer between your system and the disks. This kind of layer not only adds a
new level of complexity to your system and to your storage management, but it also has some
distance limitations between the two sites.
The second solution is based on a disk subsystem mirror such as the DS8000 Metro Mirror.
When you use a disk subsystem mirror, the synchronization process is supported by the
resources of your disk subsystem and therefore is not impacting the resources of your cluster
server nodes. At the same time, you can take advantage of the replication technology
installed on the disk subsystems, which allows replication at greater distances. It is a way to
let every actor in this clustering solution do its job. The cluster node is computing while the
disk subsystem is replicating.
586 IBM System Storage DS8000 Series: Copy Services in Open Environments
When you choose the second solution, the challenge is to automate the failover and failback
process between remote sites involving not only cluster nodes but also disk subsystems. With
IBM TotalStorage Continuous Availability for Windows you can perform such tasks
transparently from your cluster management interface.
If you choose the Active/Active cluster architecture, it means that every node at both sites can
be actively running applications. There are active resources groups on sites A and B. Your
production is naturally balanced over both nodes sharing these resource groups. When a
resource group on site A has a failure, the data and applications bound to this resource group
automatically fail over to site B. Depending on your application scripts, it can take a few
seconds or several minutes. This architecture has two advantages. First, you can be sure that
each node hosting a resource group is in a good state, and therefore not discover a wrong
behavior when it is too late (such as during a cluster failover following a disaster). Second,
you can balance your production’s workload on both nodes and therefore get better
performance. You can encounter some performance issues during a disaster scenario if the
remaining node is not able to handle the full production workload.
If you choose the Active/Passive cluster architecture, when the primary site suffers a failure,
the data and applications automatically fail over to the backup site, typically in minutes. Once
the primary site is repaired, the resources can be failed back to the original site.
For a cluster, a failover process is when the production is moving from a site A to a site B.
A failback is when the production is moving back from a site B to a site A.
For Metro Mirror, a failover is when the replication process is stopped and the production
volume is moved from site A to site B. At that time both volumes participating in the former
pair are accessible. A failback is when after a failover the replication process is reactivated
with the source volume on site B and the target volume on site A.
tape
3 Automation for Servers: z/OS, UNIX, Linux, Windows and
Business value
Autonomic
heterogeneous environments.
Services
Because the ESS 800, DS6000, and DS8000 share the same Copy Services functions you
can mix them in this solution, ESS 800 or DS6000 or DS8000. The disk subsystems are
connected to each other using PPRC links, and the host servers are fibre attached to the disk
subsystems at each site. The cluster-shared disks are mirrored between two disk
subsystems. The solution uses PPRC to coordinate the ownership and movement of disk
resources between sites.
37.3 Architecture
IBM TotalStorage Continuous Availability for Windows is a 2-site architecture with two MSCS
cluster nodes. Each client of the cluster’s resources are connected to the cluster through an
Ethernet LAN or WAN connection. For availability reasons, at least one other Ethernet
connection is linking both nodes for heartbeat purposes. The reason for this double TCP/IP
connection is that if any of this network should fall, each node could use the secondary
network to communicate, and therefore avoid an unnecessary failover.
Replication PPRC links must be defined in both directions in order for the replication pairs of
volumes to be able to fail over and fail back automatically and attempt to always replicate the
writes on the other site.
A quorum volume should be created on each site and a Metro Mirror relation should be
established between each of them. Then each volume composing a cluster resource group
should be created on both sites and a Metro Mirror relationship should be established
between each of them. Finally, for the IBM TotalStorage Continuous Availability for Windows
solution to be cluster aware, two pairs of so-called site disks should be created, and a cross
588 IBM System Storage DS8000 Series: Copy Services in Open Environments
Metro Mirror relation should be established between each of them. It means that one pair will
replicate from site A to site B and the other pair from site B to site A. These Metro Mirror
relations will never be switched in the future.
Additional scalability of extending the current 2-node cluster configuration to 4 and 8 nodes
(the maximum number of cluster nodes presently supported by Windows 2003 Enterprise
Edition) is planned for a future TotalStorage Continuous Availability for Windows release.
590 IBM System Storage DS8000 Series: Copy Services in Open Environments
A
We explain how to bring FlashCopy target volumes online to the same host as well as to a
second host. This appendix covers various UNIX and Windows platforms.
For AIX LVM, it is currently not possible to activate a Volume Group with a physical volume
(hdisk or vpath) that contains a VGID and a PVID that is already used in a Volume Group
existing on the same server. The restriction still applies even if the hdisk PVID is cleared and
reassigned with the two commands listed in Example A-1.
Therefore, it is necessary to redefine the Volume Group information about the FlashCopy
target volumes using special procedures or the recreatevg command. This will alter the
PVIDs and VGIDs in all the VGDAs of the FlashCopy target volumes, so that there are no
conflicts with existing PVIDs and VGIDs on existing Volume Groups that reside on the source
volumes. If you do not redefine the Volume Group information prior to importing the Volume
Group, then the importvg command will fail.
If the host is using LVM or MPIO definitions that work with hdisks only, follow these steps:
1. The target volume (hdisk) is new to AIX, and therefore the Configuration Manager should
be run on the specific Fibre Channel adapter:
#cfgmgr -l <fcs#>
2. Find out which of the physical volumes is your FlashCopy target volume:
#lsdev -C |grep 2107
3. Certify that all PVIDs in all hdisks that will belong to the new Volume Group were set.
Check this information using the lspv command. If they were not set, run the following
command for each one to avoid the importvg command failing:
#chdev -l <hdisk#> -a pv=yes
4. Import the target Volume Group:
#importvg -y <volume_group_name> <hdisk#>
592 IBM System Storage DS8000 Series: Copy Services in Open Environments
5. Vary on the Volume Group (the importvg command should vary on the Volume Group):
#varyonvg <volume_group_name>
6. Verify consistency of all file systems on the FlashCopy target volumes:
#fsck -y <filesystem_name>
7. Mount all the target file systems:
#mount <filesystem_name>
If the host is using SDD that works with vpath devices, follow these steps:
1. The target volume (hdisk) is new to AIX, and therefore the Configuration Manager should
be run on all Fibre Channel adapters:
#cfgmgr -l <fcs#>
2. Configure the SDD devices:
#smitty datapath_cfgall
3. Find out which of the physical volumes is your FlashCopy target volume:
#lsdev -C | grep 2107
4. Certify that all PVIDs in all hdisks that will belong to the new Volume Group were set.
Check this information using the lspv command. If they were not set, run the following
command on each one to avoid the importvg command failing:
#chdev -l <hdisk#> -a pv=yes
5. Import the target Volume Group:
#importvg -y <volume_group_name> <hdisk#>
6. Change the ODM definitions to work with the SDD devices that will provide you with the
load balance and failover functions:
#dpovgfix <volume_group_name>
7. Vary on the Volume Group (the importvg command should vary on the Volume Group):
#varyonvg <volume_group_name>
8. Verify the consistency of all file systems on the FlashCopy target volumes:
#fsck -y <filesystem_name>
9. Mount all the target file systems:
#mount <filesystem_name>
The data is now available. You can, for example, back up the data residing on the FlashCopy
target volume to a tape device. This procedure can be run once the relationship between the
FlashCopy source and target volume is established, even if data is still being copied in the
background.
The disks containing the target volumes may have been previously defined to an AIX system,
for example, if you periodically create backups using the same set of volumes. In this case,
there are two possible scenarios:
If no Volume Group, file system, or logical volume structure changes were made, then use
procedure 1 (“Procedure 1” on page 594) to access the FlashCopy target volumes from
the target system.
If some modifications to the structure of the Volume Group were made, such as changing
the file system size or the modification of logical volumes (LV), then it is recommended to
use procedure 2 (“Procedure 2” on page 594) and not procedure 1.
Procedure 2
The procedure is:
1. Unmount all the target file systems:
#umount <target_filesystem>
2. Deactivate the target Volume Group:
#varyoffvg <target_volume_group_name>
3. Export the target Volume Group:
#exportvg <target_volume_group_name>
4. Establish the FlashCopy relationships.
5. Import the target Volume Group:
#importvg -y <target_volume_group_name> <hdisk#>
6. Perform a file system consistency check on target file systems:
#fsck -y <target_file_system_name>
7. Mount all the target file systems:
#mount <target_filesystem>
Accessing the FlashCopy target volume from the same AIX host
In this section we describe a method of accessing the FlashCopy target volume on a single
AIX host while the source volume is still active on the same server. The procedure is intended
to be used as a guide and may not cover all scenarios.
If you are using the same host to work with source and target volumes, you have to use the
recreatevg command.
The recreatevg command overcomes the problem of duplicated LVM data structures and
identifiers caused by a disk duplication process such as FlashCopy. It is used to recreate an
594 IBM System Storage DS8000 Series: Copy Services in Open Environments
AIX Volume Group (VG) on a set of target volumes that are copied from a set of source
volumes belonging to a specific VG. The command will allocate new physical volume
identifiers (PVIDs) for the member disks and a new Volume Group identifier (VGID) to the
Volume Group. The command also provides options to rename the logical volumes with a
prefix you specify, and options to rename labels to specify different mount points for file
systems.
The target volume group will be tgt_flash_vg, containing vpath4 and vpath5.
Perform these tasks to run the FlashCopy and make the target volumes available to AIX:
1. Stop all I/O activities and applications that access the FlashCopy source volumes.
2. Establish the FlashCopy pairs with the No Background Copy option selected. Use the
DS Storage Manager to establish the pairs or run a script with the DS CLI command.
3. Restart applications that access the FlashCopy source volumes.
4. The target volumes, vpath4 and vpath5, will now have the same volume group data
structures as the source volumes vpath2 and vpath3. Clear the PVIDs from the target
hdisks to allow a new Volume Group to be made:
#chdev -l vpath4 -a pv=clear
#chdev -l vpath5 -a pv=clear
The output of lspv command shows the result in Figure A-1.
5. Create the target volume group and prefix all file system path names with /backup, and
prefix all AIX logical volumes with bkup:
recreatevg -y tgt_flash_vg -L /backup -Y bkup vpath4 vpath5
You must specify the hdisk names of all disk volumes participating in the volume group.
The output from lspv shown in Figure A-2 illustrates the new volume group definition.
An extract from /etc/filesystems in Figure A-3 on page 596 shows how recreatevg
generates a new file system stanza. The file system named /prodfs in the source Volume
Group is renamed to /bkp/prodfs in the target volume group. Also, the directory
/bkp/prodfs is created. Notice also that the logical volume and JFS log logical volume have
been renamed. The remainder of the stanza is the same as the stanza for /prodfs.
6. Perform a file system consistency check for all target file systems:
#fsck -y <target_file_system_name>
7. Mount the new file systems belonging to the target volume group to make them
accessible.
If the secondary volumes were previously defined on the target AIX system as hdisks or
vpaths, but the original Volume Group was removed from the primary volumes, the old
volume group and disk definitions must be removed (exportvg and rmdev) from the target
volumes and redefined (cfgmgr) before running importvg again to get the new volume group
definitions. If this is not done first, importvg will import the volume group improperly. The
volume group data structures (PVIDs and VGID) in ODM will differ from the data structures in
the VGDAs and disk volume super blocks. The file systems will not be accessible.
If the secondary volumes that are already configured on the target AIX server are in a
Remote Mirror and Copy relationship and you do not have the “Permit read access from
target” and “Reset Reserve” options enabled (and the volumes are not in full duplex state),
after rebooting the target server, the hdisks will be configured to AIX again. In other words,
you will see each Remote Mirror and Copy secondary volume twice on the target server. The
reason for this situation is as follows: AIX knows that these physical volumes already exist
with entries in the Configuration Database (ODM). However, when the configuration manager
runs during reboot, it cannot read their PVIDs because, as Remote Mirror and Copy targets,
they are locked by the DS Copy Services server. This results in AIX causing the original
hdisks to be configured to a Defined state, and new (phantom) hdisks being configured and
placed in an Available state. This is an undesirable condition that must be remedied before
the secondary volumes can be accessed.
To access the secondary volumes, the phantom hdisks must be removed and the real or
original hdisks must be changed from a Defined state to an Available state. For example,
vpath2 through vpath5 are assigned to a volume group, tgtvg. Each of the disk volumes is
596 IBM System Storage DS8000 Series: Copy Services in Open Environments
currently participating as a secondary volume in a Remote Mirror and Copy relationship. If the
server is rebooted, four new vpaths are configured to AIX. These phantom disks, vpath6
through vpath9, appear in the output from lspv, as shown in Figure A-4.
Once the updates have been imported into the secondary AIX host’s ODM, you can establish
the Remote Mirror and Copy pair again. However, select Do not copy volume from the
Select copy Options when establishing the Remote Mirror and Copy pair. As soon as the
Remote Mirror and Copy pair has been established, immediately suspend the Remote Mirror
and Copy relationship. Because there was no write I/O to the primary volumes, both the
primary and secondary are consistent.
Now that the primary volume has been suspended, the file systems can be remounted and
write I/O resumed. Once the write I/O has been going for a while, you can reestablish the
relationship with the primary and secondary by choosing Copy out-of-sync cylinders only.
If the “Permit read access from target” and “Reset Reserve” options were selected during the
Remote Mirror and Copy copy pair establish, then it would be advisable to suspend the
primary volume (while in the full duplex state) and then perform the import learn function.
Once completed, all that is necessary would be to reestablish the copy pair by only copying
the out-of-sync cylinders.
The following example shows two systems, host1 and host2, where host1 has the primary
volume vpath5 and host2 has the secondary volume vpath16. Both systems have had their
ODMs populated with the volume group itsovg from their respective Remote Mirror and Copy
volumes and, prior to any modifications, both systems’ ODM have the same time stamp, as
shown in Figure A-5.
Volumes vpath5 and vpath16 are in the Remote Mirror and Copy duplex state, and the
volume group itsovg on host1 is updated with a new logical volume. The time stamp on the
To update the ODM on the secondary server, it is advisable to suspend the Remote Mirror
and Copy pair prior to performing the importvg -L command to avoid any conflicts from LVM
actions occurring on the primary server. Figure Figure A-7 shows the updated ODM entry on
host2.
Once the importvg -L command has completed, you can reestablish the Remote Mirror and
Copy pairs and copy only the out-of-sync cylinders.
With the LDM, you are able to create logical partitions, perform disk mounts, and create
dynamic volumes. There are five types of dynamic volumes: Simple, spanned, mirrored,
striped, and RAID 5.
On Windows NT, the information relating to the disks was stored in the Windows NT registry.
With Windows 2000 and 2003, this information is stored on the disk drive itself in a partition
called the LDM database, which is kept on the last few tracks of the disk. Each volume has its
own 128-bit Globally Unique Identifier (GUID) and belongs to a disk group. This is similar to
the concept of Physical Volume Identifier (PVID) and Volume Group in AIX. As the LDM is
stored on the physical drive itself, with Windows 2000 it is possible to move disk drives
between different computers.
598 IBM System Storage DS8000 Series: Copy Services in Open Environments
Copy Services limitations with Windows 2000 and Windows 2003
Having the drive information stored on the disk itself imposes some limitations when using
Copy Services functionality on a Windows system:
The source and target volumes must be of the same physical size. Normally the target
volume can be bigger than the source volume; with Windows, this is not the case, for two
reasons:
– The LDM database holds information relating to the size of the volume. As this is
copied from the source to the target, if the target volume is a different size from the
source, then the database information will be incorrect, and the host system will return
an exception.
– The LDM database is stored at the end of the volume. The copy process is a
track-by-track copy; unless the target is an identical size to the source, the database
will not be at the end of the target volume.
It is not possible to have the source and target FlashCopy volumes on the same Windows
system when they were created as Windows dynamic volumes. The reason is that each
dynamic volume has to have its own 128-bit GUID. As its name implies, the GUID must be
unique on one system. When you perform FlashCopy, the GUID gets copied as well, so
this means that if you tried to mount the source and target volume on the same host
system, you would have two volumes with exactly the same GUID. This is not allowed,
and you will not be able to mount the target volume.
Note: When using Windows dynamic disks, remember that to read FlashCopy targets, if
the FlashCopy pair has been rescanned or if the server reading targets has been rebooted,
FlashCopy targets will appear as foreign disks to that server. Manual intervention will be
required to re-import those disks and restore operation.
Tip: Disable the Fast-indexing option on the source disk; otherwise, operations to that
volume get cached to speed up disk access. However, this means that data is not flushed
from memory and the target disk may have copies of files/folders that were deleted from
the source system.
When performing subsequent Remote Mirror and Copy/FlashCopies to the target volume, it is
not necessary to perform a reboot, because the target volume is still known to the target
system. However, in order to detect any changes to the contents of the target volume, you
There is a Windows utility, DiskPart, that enables you to script these operations so that
FlashCopy can be carried out as part of an automated backup procedure. DiskPart can be
found at the Microsoft download site:
http://www.microsoft.com/downloads
with a search on the key word DiskPart. A description of DiskPart commands can be found at
the Web site:
http://www.microsoft.com/resources/documentation/windows/xp/all/proddocs/en-us/diskpart.mspx
After reboot, the volumes will be recognized as foreign disks, and you can proceed to import
them. Reboot of the target system on subsequent FlashCopy is not necessary until the
source volume has been further extended. When performing subsequent Remote Mirror and
Copy/FlashCopy to the target volume, it is not necessary to perform a reboot because the
target volume is still known to the target system. However, in order to detect any changes to
the contents of the target volume, you should remove the drive letter from the target volume
before doing the FlashCopy. Then, after carrying out the FlashCopy, you restore the drive
letter in order for the host it is mounted on to be able to read/write to it.
There is a Windows utility, DiskPart, that enables you to script these operations so that
FlashCopy can be carried out as part of an automated backup procedure. DiskPart can be
found at the Microsoft download site with a search for the key word DiskPart.:
http://www.microsoft.com/downloads
The DiskPart tool also provides a way to extend an existing partition into free space at the
end of the same logical drive. A description of this procedure can be found in the Microsoft
Knowledge Base, article 304736:
http://support.microsoft.com/?kbid=304736
When performing subsequent Remote Mirror and Copy/FlashCopy to the target volume, it is
not necessary to perform a reboot, because the target volume is still known to the target
system. However, in order to detect any changes to the contents of the target volume, you
should remove the drive letter from the target volume before doing the FlashCopy. Then, after
600 IBM System Storage DS8000 Series: Copy Services in Open Environments
carrying out the FlashCopy, you restore the drive letter in order for the host it is mounted on to
be able to read/write to it.
Again, we refer to the Windows utility DiskPart, which enables you to script these operations
so that FlashCopy can be carried out as part of an automated backup procedure. DiskPart
can be found at the Microsoft download site:
http://www.microsoft.com/downloads;
In Windows 2000, each storage area network (SAN) hardware vendor provided its own
proprietary set of APIs for managing their hardware. This makes it challenging to develop
uniform SAN-management software. Windows Server 2003 came out with Volume Shadow
Copy Service, because it is a general infrastructure for creating consistent point-in-time
copies of data on a volume. They are generally referred to as Shadow Copies. VSS is used
for a number of purposes such as:
Creating consistent backups of open files and applications
Creating shadow copies for shared folders
For backup, testing, and data mining
The DS8000 provides an integration with Microsoft Volume Shadow Copy Service to produce
consistent shadow copies.
The IBM System Storage DS8000 Series has FlashCopy functionality, which can integrate
with the Microsoft Virtual Shadow Copy Service. DS8000 FlashCopy enables you to create
full volume copies of data in a Storage Unit. During FlashCopy operation on DS8000, it takes
a few seconds to complete the process of establishing the FlashCopy pair and creating the
necessary control bitmaps.
Figure A-8 shows the Microsoft VSS architecture and how the software provider and
hardware provider interact through Volume Shadow Copy Services.
R
R ee qq uu ee ss ttoo rr
V oo l uu m ee S hh a d oo w
CC o p yy S S e rr v i cc e W r i t ee r ss
Apps
II//O
O
S
S oo fftt w
w aa rr ee H
H aa rr dd w
w aa rr ee
P
P rr oo vv iidd ee rr P
P rr oo vv iidd ee rr
602 IBM System Storage DS8000 Series: Copy Services in Open Environments
The following list shows the Microsoft VSS FlashCopy process:
1. The requestor notifies Microsoft VSS to prepare for shadow copy creation.
2. Microsoft VSS notifies the application-specific writer to prepare its data for making a
shadow copy.
3. The writer prepares the data for that application by completing all open transactions,
flushing of cache, and writing in-memory data to disk.
4. When the data is prepared for shadow copy, the writer notifies the VSS, and it will relay
the message to the requestor to initiate the commit copy phase.
5. VSS temporarily quiesces application I/O write requests for a few seconds and the
hardware provider will perform the FlashCopy on the Storage Unit.
6. After the completion of FlashCopy, VSS releases the quiesce, and database writes will
resume.
7. VSS queries the writers to confirm that write I/Os were successfully held during Microsoft
Volume Shadow Copy.
R equestor
Backup
App W
W riters
riters
Apps
Volum
Volume
Shadow
C opy I/O
Service
D
DS8000
S8000 IB
IBM VSS Hardw are
Provider
Provider
Provider
FlashC opy
Target Source
W in2003 W in2003
Backup Production
Server Server
Additional information
You can refer to IBM TotalStorage DS Open Application Programming Interface Reference,
GC35-0493, for more technical information and implementation:
You can refer to the following Web sites for more information about Microsoft VSS:
http://www.microsoft.com/resources/documentation/WindowsServ/2003/all/techref/en-us/w2k3tr_
vss_how.asp
http://www.microsoft.com/windowsserversystem/storage/technologies/vss/default.mspx
604 IBM System Storage DS8000 Series: Copy Services in Open Environments
It unifies storage management and provides a single interface for managing block storage
virtualization. This interface is vendor and technology neutral, independent of the layer where
virtualization is done, operation system software, RAID storage hardware, or other storage
virtualization engines. Microsoft’s Virtual Disk Server enables the management of
heterogeneous storage systems, while leveraging both the client and provider’s APIs.
Microsoft VDS also supports automatic LUN configuration, which facilitates dynamic
reconfiguration by hardware in response to load or fault handling. Microsoft VDS is a software
interface for managing block storage virtualization. VDS is a vendor-specific DLL module that
integrates the vendor-specific solution into the overall VDS architecture. This interface
module makes it possible for the DS8000 storage server to interoperate with VDS-enabled
applications that utilize the VDS API in a Windows environment. VDS hides the complexities
associated with storage implementation from applications. VDS makes query and
configuration operations common across all devices it manages.
VDS is a set of Application Programming Interfaces (API) that use two sets of providers to
manage storage devices. The built-in VDS software providers enable you to manage disks
and volumes at the operating system level. VDS hardware providers supplied by the
hardware vendor enable you to manage hardware RAID arrays. Windows Server 2003
components that work with VDS include the Disk Management Microsoft Management
Console (MMC) snap-in, the DiskPart command-line tool, and the DiskRAID command-line
tool, which is available in the Windows 2003 deployment kit.
VDS utilizes a standardized interface through a graphic user interface (GUI), at the command
prompt, or through a command line interface (CLI). Figure A-10 shows the Microsoft VDS
architecture.
Command-Line Tools:
Disk Management Storage Management
•DiskPart
MMC Snap-in Applications
•DiskRAID
Other
Disk
Subsystem
HDDs LUNs
DS8000
Hardware
Microsoft Functionalitya
Functionality
Non-Microsoft Functionality
The DS8000 interacts with the IBM VDS hardware provider to Microsoft VDS. The
implementation is based on the DS CIM Agent and Microsoft VDS, using CIM technology to
query storage subsystem information and manage LUNs. Microsoft Virtual Disk Service
together with Microsoft Virtual Shadow Copy Service forms a unified heterogeneous storage
systems solution that provides:
Managing block storage virtualization
Discovery of new storage
Boot from SAN
Shadow copy creation that relates to the storage subsystem’s FlashCopy capability
Creation of consistent backups of open files and applications
Creation of shadow copies for shared folders
For backup, testing, and data mining purposes
606 IBM System Storage DS8000 Series: Copy Services in Open Environments
IBM uses QLogic SANsurfer VDS Manager to interact with Microsoft Virtual Disk Service.
Figure A-11 shows the QLogic SANsurfer VDS Manager startup screen.
Microsoft Virtual Disk Server by itself as a single component does not provide you with a
disaster recovery solution. VDS is primarily designed as a SAN management tool that allows
you to perform disk management functions. VDS provides you with an effective tool with
which you can manage multi-vendor storage systems by a single interface.
R e q u e sto r
V o lu m e S h a d o w
C o p y S e rv ice W riters
rite rs
Apps
V
VSSSS V
VDDSS I/O
S
Sooftw
ftw aare
re H
H aard
rdww aare
re H
H aard
rdww aare
re
P
Pro
rovvid
ideerr P
PPro
rovid
videerr Pro vvid
ideerr
C om m and
L in e
In te rfa ce
V irtu a l S to ra g e
D isk
is k M gt
S e rv ice
rvice App
D isk M g t
608 IBM System Storage DS8000 Series: Copy Services in Open Environments
Figure A-13 shows the Microsoft VDS and VSS architecture combined with FlashCopy.
Command
Command
Requestor Line
Line
Backup Interface
Interface
App Writers
Writers
Virtual Storage
Storage
Disk Mgt
Apps Service App
App
Volume
Shadow
Copy I/O Disk
Disk Mgt
Mgt
Service
DS8000
DS8000 IBM VSS Hardware IBM VDS
Provider Provider
Provider
FlashCopy
Target Source
Win2003 Win2003
Backup Production
Server Server
Figure A-13 Microsoft VDS and VSS with DS8000 storage subsystem
Additional information
You can refer to IBM TotalStorage DS Open Application Programming Interface Reference,
GC35-0493, for more technical information and implementation.
You can refer to the following Web site for more information about the Microsoft Virtual Disk
Service:
http://www.microsoft.com/windowsserversystem/storage/technologies/shadowcopy/stormgtusingvd
svss.mspx
The shell script to be run before the application that will use the FlashCopy target should
include the operations shown in Example A-2.
The foregoing steps for a FlashCopy operation can be done by a script, as shown in
Example A-2. This gives an example of preparing a backup using FlashCopy. Once the
FlashCopy is created the target volumes can be mounted to a different host, which can take
the data from the target volumes and send them to the final backup destination by a backup
server.
Once the Remote Mirror has been failed over or terminated, the remote host can mount the
target volumes and start the application.
610 IBM System Storage DS8000 Series: Copy Services in Open Environments
the same machine, the user needs to take some precautions. For this reason, the steps that
need to be taken are different for the two cases.
If you want to make both the source and target available to the machine at the same time, it is
necessary to change the private region of the disk, so that VERITAS Volume Manager allows
the target to be accessed as a different disk. Here we explain how to simultaneously mount
DS FlashCopy source and target volumes to the same host without exporting the source
volumes when using VERITAS Volume Manager. Check with VERITAS and IBM on the
supportability of this method before using it.
It is assumed that the sources are constantly mounted to the SUN host, the FlashCopy is
performed, and the goal is to mount the copy without unmounting the source or rebooting.
After the target volumes have been assigned, it is necessary to reboot the SUN server using
reboot -- -r or, if a reboot is not immediately possible, then issue drvconfig, disks, and
then devlinks. However, a reboot is recommended for guaranteed results.
It is also assumed that the appropriate actions in order to use the target volumes with the host
have already taken place (that is, devfsadm, vxdctl enable, and so on).
Use the following procedure to mount the targets to the same host:
1. Determine which disks have a copy of the disk group configuration in their private region.
The following command will list the log disk disks:
# vxdg list <disk group>
2. Determine the location of the private region (tag 15) on the disks (normally partition 3):
# prtvtoc /dev/rdsk/c#t#d#s2
Or use the following command to get the partition number for the private region:
# vxdisk list c#t#d#s2 | grep priv
3. Dump the private region:
# /usr/lib/vxvm/diag.d/vxprivutil dumpconfig /dev/rdsk/c#t#d#s3 > dg.dump
4. Create a script to initialize the disk group:
# cat dg.dump | vxprint -D - -d -F "vxdg -g <mydg> adddisk %name=%last_da_name" > dg.sh
5. Edit the file dg.sh and change the first line to:
# vxdg init <mydg> <daname>=<last_daname>
6. Make the file dg.sh executable:
# chmod 755 dh.sh
7. Create a file that can be used to rebuild the VM config:
# cat dg.dump | vxprint -D - -hvpsm > dg.maker
8. Initialize the disk group by executing dg.sh:
# ./dh.sh
9. If this results in the error Disk is already in use by another system, then the private
region on each disk that is to be added to the disk group will need to be initialized. This
can be done with the following command:
# vxdisksetup -i <da_name>
10. Rebuild the VM configuration:
# vxmake -g <mydg> -d dg.maker
11. Start the volumes:
# vxvol -g <mydg> start <volume>
After the secondary volumes have been assigned, it is necessary to reboot the SUN server
using reboot -- -r or, if a reboot is not immediately possible, then issue drvconfig, disks
and then devlinks. However, a reboot is recommended for guaranteed results.
Use the following procedure to mount the secondary volumes to another host:
1. Scan devices in the operating system device tree:
#vxdisk scandisks
612 IBM System Storage DS8000 Series: Copy Services in Open Environments
2. List all known disk groups on the system:
#vxdisk -o alldgs list
3. Import the Remote Mirror and Copy disk group information:
#vxdg -C import <disk_group_name>
4. Check the status of volumes in all disk groups:
#vxprint -Ath
5. Bring the disk group online:
#vxvol -g <disk_group_name> startall
or
#vxrecover -g <disk_group_name> -sb
6. Perform a consistency check on the file systems in the disk group:
#fsck -V vxfs /dev/vx/dsk/<disk_group_name>/<volume_name>
7. Mount the file system for use:
#mount -V vxfs /dev/vx/dsk/<disk_group_name>/<volume_name> /<mount_point>
Once you have finished with the Remote Mirror and Copy secondary volume, it is
recommended that you perform the following tasks:
1. Unmount the file systems in the disk group:
#umount /<mount_point>
2. Take the volumes in the disk group offline:
#vxvol -g <disk_group_name> stopall
3. Export disk group information from the system:
#vxdg deport <disk_group_name>
Tip: If you FlashCopy or Remote Mirror and Copy only one half of a RAID 1 mirror, it will be
necessary to force the import of the disk group because not all of the disks are available.
Therefore, it is necessary to issue the following command:
vxdg -f import <disk_group>
This procedure must be repeated each time you perform a FlashCopy and want to use the
target physical volume on the same host where the FlashCopy source volumes are present in
the Logical Volume Manager configuration.
3. If the target volume group exists, remove it using the vgexport command. The target
volumes cannot be members of a Volume Group when the vgimport command is run:
#vgexport -m /dev/null /dev/<target_vg_name>
4. Shut down or quiesce any applications that are accessing the FlashCopy source.
FlashCopy execution
To execute the procedure, you must carry out the following steps:
1. Unmount all file systems in the source volume group.
2. Perform the FlashCopy using the option -wait
3. Mount all the file systems in the source volume group.
4. When the FlashCopy is finished, change the Volume Group ID on each DS Volume in the
FlashCopy target. The volume ID for each volume in the FlashCopy target volume group
must be modified on the same command line. Failure to do this will result in a mismatch of
Volume Group IDs within the Volume Group. The only way to resolve this issue is to
perform the FlashCopy again and reassign the Volume Group IDs using the same
command line:
vgchgid -f </dev/rdsk/c#t#d#_1>...</dev/rdsk/c#t#d#_n>
Note: This step is not needed if another host is used to access the target devices.
614 IBM System Storage DS8000 Series: Copy Services in Open Environments
9. If the logical volume contains a VxFS file system, mount the target logical volumes on the
server:
#mount -F vxfs /dev/<target_vg_name>/<logical volume name><mount point>
Once access to the FlashCopy target volume is no longer required, unmount the file systems
and vary off the volume group:
#vgchange -a n /dev/<target_vg_name>
If no changes are made to the source volume group prior to the subsequent FlashCopy, then
all that is needed is to vary on the volume group and perform a full file system consistency
check, as shown in steps 7 to 9.
If changes are made to the source volume group, they should be reflected in the /etc/lvmtab
of the target server. Therefore, it is recommended that periodic updates be made to make the
lvmtab on both source and target machines consistent. As with the AIX importvg, there are
two alternatives:
Using the “Permit read access from target” option:
a. If you are using Global Copy, issue go-to-sync to allow the volumes to go to duplex
state.
b. Once the volumes are in the duplex state, suspend the primary volume so that no
updates are reflected on the secondary volumes.
c. Export the source volume group information into a map file.
When using Copy Services with the guest operating systems, the restrictions of the guest
operating system still apply. For example, there are some restrictions when using Copy
Services with Microsoft Windows dynamic disks, as discussed in “Windows and Remote
Mirror and Copy” on page 598. These restrictions still apply when the guest operating system
is Windows on VMware. In some cases, using Copy Services in a VMware environment may
impose additional restrictions.
Before using these techniques, check with IBM and the IBM System Storage DS8000 Host
Systems Attachment Guide, SC26-7917, for the latest information about the support available
for these solutions.
616 IBM System Storage DS8000 Series: Copy Services in Open Environments
VMware ESX server and FlashCopy
FlashCopy can be used with either the VMware Console operating system or with guest
operating systems. In general, though, FlashCopy is used when a large amount of data needs
to be copied in a very short time. The VMware Console operating system is unlikely to have
huge files (the virtual machine files tend to be quite small), so FlashCopy is not commonly
used to copy virtual machine files. Instead, we focus on using FlashCopy with VMware guest
operating systems. This focus includes the use of virtual disk files stored on VMFS disks.
Recall that VMware guest operating systems can use DS8000 volumes in one of three ways:
Virtual disk
– The volume is formatted with the VMFS file system.
– .vmdk files are created and presented to the guest operating system as a virtual disk.
– Maximizes virtualization capabilities, including a redo log.
Raw physical disk
– The volume can be directly accessed as a System LUN in physical compatibility mode.
– In this mode, I/Os are passed through the virtualization layer to the DS8000 with
minimal modifications.
– Maximizes performance.
Raw device mapping (RDM)
– The volume can be accessed as a System LUN in virtual compatibility mode.
– This mode combines the advantages of virtual disks and raw physical disks, allowing
relatively fast performance while maintaining the virtualization features of virtual disks,
such as a redo log.
Tip: When using virtual disks or raw device mappings with FlashCopy, take care which
type of Disk Mode you choose. Persistent and Nonpersistent are the easiest to manage. If
you want to use Undoable or Append, you will have to carefully manage the metadata and
be sure to copy it to the target virtual machine at the same time that you take the
FlashCopy.
Regardless of the type of virtual disk used, FlashCopy works in much the same way. This
section illustrates a method of using FlashCopy with each of these virtual disk types.
Restriction: If you do not shut down the target virtual machine first, your FlashCopied data
may not be accessible by the target virtual machine, and you may corrupt the FlashCopy
target volumes.
If you do not want to create the equivalent configuration prior to using FlashCopy, you can
instead take the following steps to get the target virtual machine to use the FlashCopy target
volumes.
Next, we make a FlashCopy of all three volumes, and then add all three volumes to the target
virtual machine. In the target virtual machine properties, choose the Hardware tab and then
Add Device. After clicking Add Device, choose Hard Disk. You will then have the choice of
Blank, Existing, or System LUN (as seen in Figure A-15).
For virtual disks located on a VMFS volume, choose Existing. For the other two types of raw
disks, use System LUN/Disk. On the subsequent panels, you should choose the same
parameters that were chosen for the source machine.
618 IBM System Storage DS8000 Series: Copy Services in Open Environments
Restriction: If your source and target virtual machines are located on the same physical
ESX Server machine, VMware will prevent you from using FlashCopy with virtual disk
located on a VMFS volume. System LUNs (both in physical compatibility mode and virtual
compatibility mode) do not have this limitation.
Now, the FlashCopy target volumes can be used as normal on the target machine.
Figure A-17 FlashCopy within the same virtual machine - After FlashCopy
620 IBM System Storage DS8000 Series: Copy Services in Open Environments
FlashCopy between virtual machines
In this scenario, we will FlashCopy two LUNs from a virtual machine, and then use the target
volumes on a second virtual machine. Figure A-18 shows the two LUNs, mounted as the F:
drive and the G: drive on the guest operating system. As above, the F: drive is a raw System
LUN in physical compatibility mode, while the G: drive is a raw System LUN in virtual
compatibility mode. Both drives contain a directory called VM1.
Before issuing the FlashCopy, it is important to prepare both the source and target machines
to be copied. For the source machine, this typically means quiescing the applications,
unmounting the source volumes, and/or flushing memory buffers to disk. See the appropriate
sections for your operating systems for more information about this topic.
For the target machine, typically the target volumes must be unmounted. This prevents the
operating system from accidentally corrupting the target volumes with buffered writes, as well
as preventing users from accessing the target LUNs until the FlashCopy is logically complete.
With VMware, there is an additional restriction that the target virtual machine must be shut
down before issuing the FlashCopy. VMware also performs caching, in addition to any
caching the guest operating system might do.
622 IBM System Storage DS8000 Series: Copy Services in Open Environments
After preparing both the source and target machines, issue the FlashCopy and then start your
target virtual machine. Figure A-20 shows that the second virtual machine’s files have been
replaced.
Figure A-20 FlashCopy between virtual machines - Target virtual machine after FlashCopy
On the target virtual machine on a second ESX Server machine, as seen in Figure A-22 on
page 625, the LUNs have been configured and assigned. Disk 1 is the target’s virtual disk on
a VMFS volume, while Disk 2 is a raw System LUN in physical compatibility mode, and Disk 3
is a raw System LUN in virtual compatibility mode. No Windows file systems have been
created, and no drive letters have been assigned to these volumes.
Again, in order to obtain a consistent copy of the source LUNs, you have to prepare both the
source and target virtual machines. This means quiescing the source volumes like you would
do when FlashCopying native Windows or Linux volumes. This also means that the target
virtual machine should be powered off. After these preparations, you are ready to issue the
FlashCopy command.
624 IBM System Storage DS8000 Series: Copy Services in Open Environments
In Figure A-22, you can see that the target volumes have been assigned to the target virtual
machines, but no volumes have been created.
Figure A-23 FlashCopy between ESX Servers - Target virtual machine after FlashCopy
On subsequent FlashCopies, you may not need to mount the volumes or assign drive letters
again. Unless you reconfigure, the target virtual machine will remember the previous mount
points.
As with FlashCopy, using VMware with Remote Mirror and Copy contains all the advantages
and limitations of the guest operating system. See the individual guest operating system
sections for relevant information. Using VMware with Remote Mirror and Copy imposes some
additional restrictions.
One such limitation is the mkpprc -tgtread parameter is not supported. VMware cannot use
VMFS-formatted volumes or raw System LUNs in virtual mode without writing to the disk.
However, it may be possible to use raw System LUNs in physical compatibility mode. Check
with IBM on the supportability of this procedure.
At a high level, the steps for creating a Remote Mirror and Copy are as follows:
1. Shut down the guest operating system on the target ESX Server.
626 IBM System Storage DS8000 Series: Copy Services in Open Environments
2. Establish remote mirror and copy from the source volumes to the target volumes.
Important: You should use the mkpprc -resetreserve flag when establishing the
Remote Mirror and Copy. Otherwise, you may receive this message on the target
machine:
Cannot create partition table for disk vmhbax:x:x because geometry info is invalid.
Please rescan
This flag removes hardware protection on the target device and should be used with
caution.
3. When the initial copy has completed and the volumes are now in Full Duplex mode,
suspend or remove the Remote Mirror and Copy relationship.
4. Issue the Rescan SAN command on the target ESX Server.
5. If not already assigned to the target virtual machine, assign the Remote Mirror and Copy
volumes to the target virtual machine. VMFS-formatted volumes should be assigned as
existing volumes, while raw volumes should be assigned as System LUNs of the same
type as the source.
6. Start the virtual machine and, if necessary, mount the target volumes.
The family of servers described in this chapter was introduced in 1988 as AS/400. In the
mid-nineties, the brand name was changed to iSeries. The name changed again recently to
System i or System i5. System i refers to entire product line, including older models, while
System i5 refers only to models based on the Power5 processor.
In this chapter we use the name System i5 because some of the solutions we discuss use
Boot from SAN and this function can only be realized on System i5. Note that other solutions
discussed here rely on Independent Auxiliary Storage Pools (IASPs), and these can be
implemented on older models as well.
Only in some instances when we want to point out a characteristic that apply to all models, do
we use the name System i.
Today, System i5 clients have a variety of choices for external storage solutions and features.
With the DS6000 or DS8000 storage systems, they can use Metro Mirror or Global Mirror for
disaster recovery, they can replicate all disks or just application data, and they can use
FlashCopy to minimize downtime when taking backups.
This chapter presents and describes solutions to minimize downtime. First we review the
System i5 functions that are relevant to implementing external storage. An understanding of
those functions is critical to properly propose and manage external storage based solutions
and scenarios.
When discussing business continuity solutions for a System i5 audience, we use terms such
as continuous operations, high availability and continuous availability, which are usually
used within the System i community. We have included a formal definition in the note below.
Definitions:
Continuous Operations (CO): Attribute of a system ability to continuously operate and
mask planned outages from end-users. It employs Non-disruptive hardware and
software changes, nondisruptive configuration, software coexistence.
High Availability (HA): Attribute of a system ability to provide service during defined
periods, at acceptable or agreed upon levels, and mask unplanned outages from end
users. It employs fault tolerance, automated failure detection, recovery, bypass
reconfiguration, testing, and problem and change management
Continuous Availability (CA): Attribute of a system ability to deliver non disruptive
service to the end user 7 days a week, 24 hours a day (there are no planned or
unplanned outages).
630 IBM System Storage DS8000 Series: Copy Services in Open Environments
System i5 structure
System i5 is the newest generation of System i (formerly iSeries) servers. It is based on the
POWER5 processor. A System i5 can run one or multiple logical partitions, managed by an
hypervisor that resides in main memory.
A partition can host one of the following operating systems: i5/OS (formerly OS/400®), Linux,
or AIX.
Configure and manage partitions through a Hardware Management Console (HMC) that is
connected to the System i5 via an Ethernet connection.
SLIC SLIC
Firmware
Private
Desktop
Network
OR Firmware
and/or Perm | Temp
Public
Network
Rack mount
In this chapter we discuss solutions with i5/OS partitions and DS systems. In the remaining of
this chapter, we refer to an i5/OS partition in System i5, simply as partition.
Single-level storage
For storage, the System i5 with i5/OS uses the same architectural component already used
by iSeries and AS/400, that is a single-level storage. This means that the System i5 sees all
disk space and the main memory as one storage area, and uses the same set of virtual
addresses to cover both main memory and disk space.
Main Memory
When the application performs an input output (IO) operation, the portion of the program that
contains read or write instructions is first brought into main memory where the instructions are
then executing.
With the read request, the virtual addresses of the needed record are resolved and for each
needed page, storage management first looks to see if it is in the main memory. If the page is
there, it is used for resolving the read request. But if the corresponding page is not in the main
memory, it must be retrieved from disk (page fault). When a page is retrieved, it replaces a
page that was recently not used; the replaced page is swapped to disk.
Similarly, writing a new record or updating an existing record is done in main memory, and
the affected pages are marked as changed. A changed page remains in main memory until it
is swapped to disk as a result of a page fault. Pages are also written to disk when a file is
closed or when write to disk is forced by a user through commands and parameters. Also,
database journals are written to the disk.
A subject in System i5 is anything that exists and occupies space in storage and on which
operations can be performed. For example, a library, a database file, a user profile, a
program, are objects in System i5.
632 IBM System Storage DS8000 Series: Copy Services in Open Environments
IOP is to provide a certain degree of managing IO operations, and consequently off load
some of the I/O-related work from the System i5 central processor. Fibre Channel (FC)
adapters are used to connect external storage, attach to IOPs, and I/O operations to and from
external storage are managed by IOPs.
Some of the functions performed by an IOP with attached FC adapter include translating a
request for a block of data so that the adapter can understand it and informing the adapter of
the location of a data block in main memory.
System i5 Partition
Main memory
RIO
IOP IOP
PCI-X PCI-X
FC FC
ad ad
SAN
SAN
DS - External Storage
Clusters
Many of System i5 continuous availability solutions which use external storage are based on
clusters and independent disk pools (independent auxiliary storage pools).
Clusters provide continuous availability in several ways: they use techniques like switched
disks, replicating data with Cross Site Mirroring, or replicating data with external disks while
using the copy services functions they provide. Continuous availability solutions from ISVs
also use System i5 clusters.
A System i5 cluster is a group of one or more systems or logical partitions that work together
as a single system.
CRG
C
CRG CRG CRG
B B C
Recovery Cluster
Domain Resource
Group
Cluster Resources
(e.g., IASP)
Once the cluster is set up, the necessary information is maintained on all cluster nodes
through the heartbeating which runs on IP connection among the nodes.
634 IBM System Storage DS8000 Series: Copy Services in Open Environments
If a system outage or a site loss occurs, the functions that are provided on a system or
partition within a cluster can be accessed through other systems or partitions that have been
defined in the cluster. When maintenance is needed on production partition, another node in
a cluster can handle resources of production partition and continues production work. This
functionality is achieved through cluster events like failover, switchover, replication, and
rejoin.
Solutions that are based on copy services provided by external storage systems and
Independent Auxiliary Storage Pool (IASP) require that all involved systems are in a cluster.
For more information about System i5 clusters refer to the iSeries Information Center on the
following Web page:
http://publib.boulder.ibm.com/iseries/
Also refer to the IBM Redbook Clustering and IASPs for Higher Availability on the
IBM Eserver iSeries Server, SG24-5194.
However, for many years, customers have found needs for additional storage granularity,
including the need to sometimes isolate data into a separate disk pool. Since the first AS/400,
many customers have utilized User ASPs to isolate disk units for journaling data (logging),
saving data to disk, and isolating slower, denser, less costly disk units for archive needs. User
ASP provide the same automation and ease-of-use benefits as the System ASP, but provide
additional storage isolation when needed. With software level Version 5, IBM takes this
storage granularity option a huge step forward with the availability of Independent Auxiliary
Storage Pools (IASPs).
An IASP is a collection of disk units that can be brought online or taken off-line independent of
the rest of the storage on a system. An IASP can be switched between systems or logical
partitions. And unlike User ASPs, IASPs coupled with i5/OS Clusters allow granular
replication based on the IASP, using storage options such as i5/OS Cross Site Mirroring or
IBM Storage Systems Copy Services.
IASPs represent a strategic direction for System i Storage Management. IASPs help keep the
System i promise of simplified IT management, while joining together the need for more
storage granularity, flexibility, and storage options
A system or partition has always a system ASP where vital system data reside. It can also
have user ASPs and IASPs. In partition with IASPs, we use the term sysbas. Sysbas refers to
system ASP plus user ASPs. When talking about solutions in such partition we usually
distinguish between IASPs and sysbas.
IASPs can be on internal disks in System i5 (can contain internal disks), or they can be on
external storage (contain volumes from ESS, DS6000, or DS8000).
When an IASP resides on external storage it can be replicated by DS6000 or DS8000 Copy
Services. The copy of an IASP can then be varied on to another System i or partition in a
cluster, making applications and data in that IASP available to the other system.
Continuous availability solutions for IASPs residing on DS6000 or DS8000 external storage,
are based on Metro Mirror or Global Mirror of IASP volumes. Solutions for minimizing
downtime for backups are based on FlashCopy of IASP volumes.
For more information about Metro Mirror refer to Part 4, “Metro Mirror” on page 107. For more
information about IASPs refer to “Independent Auxiliary Storage Pools (IASPs)” on page 635.
When describing functions of solutions with DS Copy services, we use terms planned outage,
unplanned outage and switchover. They are formally defined below.
636 IBM System Storage DS8000 Series: Copy Services in Open Environments
Note: During a planned outage (also called a scheduled outage), you deliberately make
your system unavailable to users. You might use a scheduled outage to run batch work,
back up your server, or apply fixes.
A switchover is a manual process initiated by the user in order to move server functions
during a planned outage of the primary server to the backup server. This process involves
the quiescing of the application environment and system functions on the primary server
and the initiation of the application and system resource on the backup system. A
switchover can take anywhere from minutes to hours depending on the complexities of the
infrastructure and application environment.
Solution description
Our setup scenario for this solution consists of a local partition and a remote partition in
separate System i5 servers, but both partitions are grouped in a cluster.
Critical application data reside in an IASP in the local partition, and the IASP contains
volumes that reside on an external disk system. A Metro Mirror relationship is established
between the volumes in that IASP and volumes in another, remote disk system to which the
remote partition has access.
The Metro Mirror secondary volumes (in other words, the exact copy of the production IASP)
can be varied on to the recovery partition.
Note that it is only necessary that IASP on both partitions reside on external storage. sysbas
in each partition can be on internal disks or on external disk system. For more information
about sysbas refer to “Independent Auxiliary Storage Pools (IASPs)” on page 635.
Metro IASP
Mirror
IASP
This solution provides continuous availability in case of planned and unplanned outages:
For each planned or unplanned outage, the Metro Mirror copy of the production IASP is
made available to the remote partition which continues to run production application from
the data in the mirrored IASP.
When the planned outage is finished or the cause of failure of the unplanned outage has
been fixed, a reverse of the Metro Mirror direction (from secondary volumes to primary
volumes) is performed to transfer updated data in the remote IASP back to the local
partition. When the transfer is complete, the Metro Mirror relationship is established again
in the original direction (from the primary to the secondary volume) and production can
resume at the primary site.
This solution is implemented with the System i Copy Services Toolkit available for purchase
from the System i Client Technology Center in Rochester.
Note: The System i Copy Services Toolkit is a software product that gets installed on all
partitions in a FlashCopy or PPRC clustered environment.
Solution benefits
Metro Mirror of an IASP offers several benefits:
Basically, recovery takes as long as is needed to vary on an IASP. Therefore recovery
time is significantly shorter than with solutions which use Boot from SAN, or external
loadsource, where IPL is needed for recovery. Recovery time with Metro Mirror of an IASP
can be compared to the recovery time for high availability solutions provided by ISVs.
Note that Boot from SAN is an IBM implementation of an external boot disk in System i. In
this implementation a boot disk is connected through a special input output processor
638 IBM System Storage DS8000 Series: Copy Services in Open Environments
(IOP) and an FC adapter in System i5. For more information about Boot from SAN refer to
the IBM Redbook IBM System Storage DS6000 Series: Architecture and Implementation,
SG24-6781. External loadsource (boot disk) is also implemented by competitor’s external
storage solutions, though it doesn’t use Boot from SAN IOP.
Due to synchronous copying of writes, Metro Mirror has always some impact on
production performance. With this solution we only use Metro Mirror of data in an IASP.
Therefore impact on performance is relatively small. On the other hand, solutions with
Boot from SAN or external loadsource replicate the entire partition disk space, including
temporary files, job queues, and so on and the performance degradation of the production
system can be more significant.
This solution only has impact on write operations performances. It does not use any of the
System i5 CPU resources. This is different from some high availability implementations
provided by other vendors, where there is an impact to performances due to CPU usage
as part of their solution.
With this solution we can run a non production workload, like test or development tasks in
the remote partition. This is again different from the implementation using Boot from SAN
where the recovery parititon has to be in stand-by, typically without any workload.
The System i Copy Services Toolkit which is used for this solution provides full automation
of steps performed at planned or unplanned outages. All the needed steps for failover and
failback are initiated by one command in a partition of System i5.
Once this solution is established it requires very little maintenance and administration.
Therefore some customers may prefer it to other high availability solutions available in
System i5.
Note: For information about the Toolkit contact the Client Technology Center, via the
following Internet page:
http://www-03.ibm.com/servers/eserver/services/iseriesservices.html
Solutions that use the System i Copy Services Toolkit cannot be sold to the customer
without prior approval by the Client Technology Center.
The solution requires a System i and a DS6000, DS8000, or ESS at both local and remote
sites. It also requires appropriate FC connections between the DS systems for Metro
Mirror and connections between Systems i5 for cluster communication via TCP/IP.
Each IASP has its own Fibre Channel attachment cards. This is valid for both System i5
and earlier System i models like 8xx and 270.
A Metro Mirror license on both local and remote disk systems is required.
i5/OS software for clustering (5722-SS1 option 41) must be installed in both production
and recovery partition. This prerequisite is needed on each System i5 and earlier System i
models like 8xx and 270.
Considerations
Following points should be considered with this solution:
An application in System i5 typically consists of different objects, like programs, data files,
data areas, spool files user profiles and others. After the application is set up in an IASP,
the majority of application objects reside in the IASP. However, some objects may reside
in sysbas. For example, data files that belong to the application go to the IASP while the
programs and job queues may stay in sysbas.
It is necessary to maintain application objects in sysbas on both production and recovery
partitions in sync, to allow the application to run in the recovery parititon after Metro Mirror
copy of the production IASP is made available to it.
As experienced with many applications, maintenance of sysbas is rather simple and
straightforward: it is achieved by i5/OS change control, or by installing new application
versions on both partitions at the same time. However, some customers may want to use
other ways of maintaining sysbas, like regularly saving the objects in the production
partition and restoring them in the recovery partition, or using a limited version of an ISV
solution.
For more information about i5/OS change control refer to the iSeries Information Center
on the following Web page and look for Management Central:
http://publib.boulder.ibm.com/infocenter/iseries/v5r4/index.jsp
640 IBM System Storage DS8000 Series: Copy Services in Open Environments
i5/OS release V5R4 brings significant improvement in replicating i5/OS user profiles that
reside in sysbas. With this release it is possible to define a new type of cluster domain,
administrative domain, which enables that changes in user profiles are automatically
propagated from the local sysbas to remote sysbas. For more information about cluster
domains refer to “Clusters” on page 633.
We strongly recommend journaling the application objects when using Metro Mirror for an
IASP, or even Metro Mirror of an entire disk space.
Important: When using Copy Services functions such as Metro Mirror, Global Mirror, or
FlashCopy for the replication of the load source unit or other i5/OS disk units within the
same DS6000/DS8000 or between two or more DS6000 and DS8000 systems, the source
volume and the target volume characteristics must be identical. The target and source
must be of matching capacities and matching protection types.
Additionally, once a volume is assigned to an i5/OS partition and added to that partition’s
configuration, its characteristics must not be changed. If there is a requirement to change
some characteristic of a configured volume, it must first be completely removed from the
i5/OS configuration. After the characteristic changes are made, for example, protection
type, capacity, and so on, by destroying and recreating the volume or by utilizing the DS
CLI, the volume can be reassigned to the i5/OS configuration. To simplify the
configuration, we recommend a symmetrical configuration between two IBM Storage
systems, creating the same volumes with the same volume ID that decides the LSS_ID.
After the toolkit is installed and tested you can perform switchover and failback during
planned and unplanned outages by using toolkit commands. The following toolkit commands
are used:
swpprc - You use this command with parameters *scheduled or *unscheduled depending
on whether a planned or unplanned outage is being handled. The swpprc command
initiates a sequence of steps in the Toolkit by which the required actions (switchover,
failback, and other actions) are achieved.
You can also check whether the solution is in the correct status by using the System i5
chkpprc and dspessdta commands from the toolkit.
When such a System i5 command is executed, the Toolkit triggers the different tasks by
communicating with the System i5 HMC, performing System i5 commands within the
partition, and communicating with external disks systems by using the DSCLI. This is
illustrated in Figure B-7.
SSH HMC
HMC
TCP/IP SSH
TCP/IP
DS/ ESS
Toolkit Toolkit
FSP FSP
IASP IASP’
DS CLI DS CLI
(FIBRE)
(FIBRE)
IOA IOA
Below we describe the sequence of steps required for a switchover, for checking the Metro
Mirror relationship status, and for setting up the toolkit.
Note that in the examples shown below, the partition ITCHA2 is a local parititon, ITCHA3 is a
remote partition, and the IASP name is DS6000.
642 IBM System Storage DS8000 Series: Copy Services in Open Environments
Checking IASP status
To observe the logical volumes in the IASP, use the System i5 graphical interface known as
iSeries Navigator, which is a part of the iSeries Access for Windows licensed product.
Instructions on how to install iSeries Access for Windows can be found in the iSeries
Information Center located at:
http://publib.boulder.ibm.com/infocenter/iseries/v5r4/index.jsp
In iSeries Navigator make a connection to the System i5 that contains the IASP.
Note: Thew System i5 user ID and password are required to connect to the System i5.
After the connection is established, look for the system in the left-hand side panel of the
iSeries Navigator window and expand it. Then expand Configuration and Service →
Hardware → Disk Units → Disk Pools.
Note that for expanding disk units you will need a System i5 Service Tools user ID and
password.
Observe that IASPs are numbered 33 and higher, while traditional disk pools (user ASPs)
have numbers below 33. Click the relevant IASP to show its disks in the right-hand side
panel. This is shown in Figure B-8.
To check the status of the IASP, you can also use the i5/OS wrkcfgsts command.
After entering this command, press F4, which brings up a screen where you can insert
parameters. At *type insert *dev, at Configuration description, insert the name of IASP. Leave
parameter Output as *, as shown in Figure B-9 on page 644, and press Enter twice.
Bottom
F3=Exit F4=Prompt F5=Refresh F10=Additional parameters F12=Cancel
F13=How to use this display F24=More keys
The next screen, as shown in Figure B-10. displays the name and status of the IASP.
Bottom
Parameters or command
===>
F3=Exit F4=Prompt F12=Cancel F23=More options F24=More keys
644 IBM System Storage DS8000 Series: Copy Services in Open Environments
information about how to use IBM Personal Communications to Telnet to a System i5, refer to
the iSeries Information Center located at:
http://publib.boulder.ibm.com/infocenter/iseries/v5r4/index.jsp
When the Telnet connection is established, specify the System i5 user ID and password to
sign in. Type dspessdta in the command line of the System i5 Telnet screen, as shown in
Figure B-11, and press Enter.
1. User tasks
2. Office tasks
3. General system tasks
4. Files, libraries, and folders
5. Programming
6. Communications
7. Define or change the system
8. Problem handling
9. Display a menu
10. Information Assistant options
11. iSeries Access tasks
Selection or command
===> dspessdta
A new screen now displays the IASP name and other options or status information such as
the Metro Mirror direction, auto-start of cluster after link failure, and so on. This is shown in
Figure B-13. We recommend using this command and observing the options and status
before performing a switchover. After you have checked the values displayed for the different
fields, press F3 or F12 to exit this screen and will bring back the i5/OS command line.
Bottom
F1=Help F3=Exit F12=Cancel
646 IBM System Storage DS8000 Series: Copy Services in Open Environments
Checking the Metro Mirror status
To check for a correct status of the DS storage systems prior to the switchover, use the
System i5 chkpprc command and specify the IASP name, as is shown in Figure B-14. We
recommend regularly performing the chkpprc command to be sure that DS systems are in a
correct status for switching to the remote site. You can execute this command in either the
local or recovery partition.
1. User tasks
2. Office tasks
3. General system tasks
4. Files, libraries, and folders
5. Programming
6. Communications
7. Define or change the system
8. Problem handling
9. Display a menu
10. Information Assistant options
11. iSeries Access tasks
Selection or command
===> CHKPPRC ds6000
1. User tasks
2. Office tasks
3. General system tasks
4. Files, libraries, and folders
5. Programming
6. Communications
7. Define or change the system
8. Problem handling
9. Display a menu
10. Information Assistant options
11. iSeries Access tasks
Selection or command
===> CHKPPRC ds6000
Finally you are presented the message that PPRC was successfully checked, or you are
informed about a failed check. This is shown on Figure B-16 on page 649.
648 IBM System Storage DS8000 Series: Copy Services in Open Environments
MAIN i5/OS Main Menu
System: ITCHA2
Select one of the following:
1. User tasks
2. Office tasks
3. General system tasks
4. Files, libraries, and folders
5. Programming
6. Communications
7. Define or change the system
8. Problem handling
9. Display a menu
10. Information Assistant options
11. iSeries Access tasks
Selection or command
===> CHKPPRC ds6000
1. User tasks
2. Office tasks
3. General system tasks
4. Files, libraries, and folders
5. Programming
6. Communications
7. Define or change the system
8. Problem handling
9. Display a menu
10. Information Assistant options
11. iSeries Access tasks
Selection or command
===> viewlog
650 IBM System Storage DS8000 Series: Copy Services in Open Environments
You are presented with the Toolkit log, where you can see the results of checking the Metro
Mirror status, as shown in Figure B-18.
CMD ....+....1....+....2....+....3....+....4....+....5....+....6....+....7....+
Fri Jun 23 08:01:16 2006 workWithHmcProfile: FYI, /QIBM/Qzrdiash5/hmc/DS600
"21020003/none/1,21030002/none/1,21050002/none/0,21010003/none/1,21030003/
Fri Jun 23 08:01:16 2006 doPPRCScript: Executing: QDSCLI/DSCLI SCRIPT('/QIB
Fri Jun 23 08:01:25 2006 checkThoseResults: Processing file /QIBM/Qzrdiash5
Fri Jun 23 08:01:25 2006 checkThoseResults: Strings Full Duplex -Target Ful
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
Parameter IASPNAME required.
Directory: /qibm/qzrdiash5/profiles/DS6000
Position to : Record : 1 of 4
New File :
2=Edit 4=Delete File 5=Display 6=Path Size 9=Recursive Delete
Bottom
3. Press Enter to get the screen showing the selected DS CLI profile, as shown in
Figure B-21.
Browse : /qibm/qzrdiash5/profiles/DS6000/pprc_A.profile
Record : 1 of 82 by 14 Column : 1 68 by 79
Control :
....+....1....+....2....+....3....+....4....+....5....+....6....+....7....+....
************Beginning of data**************
#
# DS CLI Profile
#
#
# Management Console/Node IP Address(es)
# hmc1 and hmc2 are equivalent to -hmc1 and -hmc2 command options.
hmc1: 9.5.110.66
#hmc2: 127.0.0.1
pwfile: /qibm/qzrdiash5/profiles/DS6000/pprc_Asec.dat
#
# Default target Storage Image ID
652 IBM System Storage DS8000 Series: Copy Services in Open Environments
Checking a DS CLI script
To check a DS CLI script in a System i5 partition, enter the viewscript command with the
IASP name as the parameter. You are presented with a screen listing the DS CLI scripts used
in this partition. Look for the script you want to examine and type 5 in the Opt field next to this
script, as is shown in Figure B-22. Press Enter.
Directory: /qibm/qzrdiash5/scripts/DS6000
Position to : Record : 1 of
New File :
2=Edit 4=Delete File 5=Display 6=Path Size 9=Recursive Delete
Browse : /qibm/qzrdiash5/scripts/DS6000/pausepprc_B.script
Record : 1 of 2 by 14 Column : 1 80 by 79
Control :
....+....1....+....2....+....3....+....4....+....5....+....6....+....7....+....
************Beginning of data**************
# Generated: Thu Jun 22 16:23:10 2006
pausepprc -dev IBM.1750-6877801 -remotedev IBM.1750-13ADLWA 1200-1204:1200-1204
************End of Data********************
1. User tasks
2. Office tasks
3. General system tasks
4. Files, libraries, and folders
5. Programming
6. Communications
7. Define or change the system
8. Problem handling
9. Display a menu
10. Information Assistant options
11. iSeries Access tasks
Selection or command
===> swpprc
654 IBM System Storage DS8000 Series: Copy Services in Open Environments
This brings up a screen where you can insert the IASP name and specify the parameter
*scheduled or *unscheduled. If you perform a switch to the production system for planned
outage specify *scheduled, if you switch to production for an unplanned outage, specify
*unscheduled.
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
The swpprc toolkit command executes all the required actions to switch over to the IASP at
the recovery site, for both systems and external storage. The following actions take place
during the execution of swpprc *scheduled:
1. Run chkpprc command.
2. Vary off the IASP on local partition.
3. Perform Metro Mirror failover to secondary site.
For more information about Metro Mirror failover refer to 12.3, “Failover and Failback” on
page 115.
4. Perform Metro Mirror failback from secondary to primary site. For more information about
Metro Mirror failback refer to 12.3, “Failover and Failback” on page 115.
5. Vary on the IASP for remote partition.
6. By using System i5 logical partitioning functions, remove IOPs and adapters logically from
the former production system and modify the description of the local System i5 partition.
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
Checking lspprc DSCLI script for production node.
656 IBM System Storage DS8000 Series: Copy Services in Open Environments
After swpprc *scheduled is performed you can use the viewlog command to check if it was
succesfuly executed. An example is shown in Figure B-27.
CMD ....+....1....+....2....+....3....+....4....+....5....+....6....+....7....+
Fri Jun 23 09:13:22 2006 workWithHmcProfile: FYI, issued command:
chsyscfg -r prof -m "ITCHA3" -i "name=ITCHA3,lpar_name=ITCHA3,io_slots=\\\
Fri Jun 23 09:14:23 2006 startDasdMgmtOperaton: First of 2 ASP records retu
Fri Jun 23 09:14:23 2006 startDasdMgmtOperaton: MultiPath has been reset fo
Fri Jun 23 09:14:23 2006 startDasdMgmtOperaton: Record 2 of 2 ASP records r
Fri Jun 23 09:14:23 2006 startDasdMgmtOperaton: MultiPath has been reset fo
Fri Jun 23 09:15:40 2006 addRemoveHmcDevices: Removed device at bus 2, slot
Fri Jun 23 09:15:40 2006 addRemoveHmcDevices: Removed a total of 1 device(s
Fri Jun 23 09:15:41 2006 workWithHmcProfile: FYI, /QIBM/Qzrdiash5/hmc/DS600
"21030002/none/0,21020003/none/1,21050002/none/0,21010003/none/1,21040002/
Fri Jun 23 09:15:41 2006 workWithHmcProfile: FYI, device on ITCHA2 bus 2 sl
Fri Jun 23 09:15:53 2006 workWithHmcProfile: FYI, issued command:
chsyscfg -r prof -m "ITCHA2" -i "name=ITCHA2,lpar_name=ITCHA2,io_slots=\\\
Fri Jun 23 09:15:53 2006 swpprc: SUCCESSFUL.
************End of Data********************
You can check the status of the IASP copy, which is now available at the remote partition.
Proceed as follows:
1. In the remote partition, enter the command wrkscfgsts cfgtype(*dev) cfgd(ds6000).
This will bring up a screen showing the IASP status. After successful switchover, the IASP
should have a status of available, as shown on Figure B-28.
Bottom
===>
F21=Select assistance level
2. Since varying off the production IASP is done as a part of swpprc, the IASP is no longer
available to the local partition after the switchover. By using the command wrkcfgsts
3. You can now observe the toolkit status and options by entering the command dspessdta
with the IASP name DS6000, in our case, as parameter. As shown in Figure B-30, the
direction of Metro Mirror is now reversed, and the current production partition is ITCHA3,
which was originally the recovery partition.
658 IBM System Storage DS8000 Series: Copy Services in Open Environments
specify how to switch back: *scheduled for planned outages or *unscheduled for unplanned
outages. Specify *scheduled, is shown in Figure B-31.
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
Parameter IASPNAME required.
While swpprc executes, messages indicating progress with the above actions appear at the
bottom of the screen.
CMD ....+....1....+....2....+....3....+....4....+....5....+....6....+....7....+
Fri Jun 23 09:38:17 2006 startDasdMgmtOperaton: First of 2 ASP records retu
Fri Jun 23 09:38:17 2006 startDasdMgmtOperaton: MultiPath has been reset fo
Fri Jun 23 09:38:17 2006 startDasdMgmtOperaton: Record 2 of 2 ASP records r
Fri Jun 23 09:38:17 2006 startDasdMgmtOperaton: MultiPath has been reset fo
Fri Jun 23 09:39:39 2006 addRemoveHmcDevices: Removed device at bus 2, slot
Fri Jun 23 09:39:39 2006 addRemoveHmcDevices: Removed a total of 1 device(s
Fri Jun 23 09:39:40 2006 workWithHmcProfile: FYI, /QIBM/Qzrdiash5/hmc/DS600
"21020003/none/1,21030002/none/1,21050002/none/0,21010003/none/1,21040002/
Fri Jun 23 09:39:40 2006 workWithHmcProfile: FYI, device on ITCHA3 bus 2 sl
Fri Jun 23 09:39:52 2006 workWithHmcProfile: FYI, issued command:
chsyscfg -r prof -m "ITCHA3" -i "name=ITCHA3,lpar_name=ITCHA3,io_slots=\\\
Fri Jun 23 09:39:52 2006 swpprc: SUCCESSFUL.
************End of Data********************
660 IBM System Storage DS8000 Series: Copy Services in Open Environments
Check the status of the production IASP by entering the command swrkcfgsts
cfgtypw(*dev) cfgd(ds6000) on the local partition. The status should be available, as shown
in Figure B-33.
Bottom
Parameters or command
===>
F3=Exit F4=Prompt F12=Cancel F23=More options F24=More keys
You can also check the toolkit status by entering the dspessdta command with the IASP name
as the parameter. As is shown in Figure B-34, the direction of the Metro Mirror is now normal.
Bottom
F1=Help F3=Exit F12=Cancel
Re-trying SWPPRC
If for any reason executing swpprc fails before completion, you must run it again with the
*retry parameter to complete the switchover or switchback process.
A unscheduled switch will always be an incomplete switch due to failures. The command
swpprc with parameter *complete must be run once the failures have been corrected. This will
complete Metro Mirror failover. These commands are described below.
Use the following steps to switch to the remote site in unplanned outage situations:
1. At the failure of production site, use the command swpprc *unscheduled in the remote
partition, as shown in Figure B-35.
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
662 IBM System Storage DS8000 Series: Copy Services in Open Environments
Shortly after the command starts executing, the message Waiting for reply to message
on message queue QSYSOPR appears, as shown in Figure B-36.
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
Waiting for reply to message on message queue QSYSOPR.
System Request
System: ITCHA3
Select one of the following:
Bottom
Selection
6
F3=Exit F12=Cancel
(C) COPYRIGHT IBM CORP. 1980, 2005.
Display Messages
System: ITCHA3
Queue . . . . . : QSYSOPR Program . . . . : *DSPMSG
Library . . . : QSYS Library . . . :
Severity . . . : 99 Delivery . . . : *HOLD
When confirming to perform the unscheduled switch, the command executes further.
During execution, messages indicating the actions performed are displayed at the bottom
of the screen. The messages are similar to the ones shown in “Switch to remote site at
planned outages” on page 654.
664 IBM System Storage DS8000 Series: Copy Services in Open Environments
3. Once the command swpprc *unscheduled has completed, use the command dspessdta
with the IASP name as the parameter to display the Metro Mirror status. Output of this
command is shown in Figure B-39. Observe that the Metro Mirror status is incomplete.
The command dspessdta is described in “Checking the Toolkit setup” on page 644.
You can also look at the toolkit log to check for successful execution of the swpprc
*unscheduled command. For information about how to view the log, refer to “Checking the
Toolkit setup” on page 644.
4. After the local site is operational again, perform the command swpprc *complete at the
local site to complete the switch. The command is shown in Figure B-40.
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
Waiting for reply to message on message queue QSYSOPR.
5. To answer the message open another telnet session with the local partition, enter the
command dspmsg qsysopr, and reply g. This is illustrated in Figure B-42.
Display Messages
System: ITCHA2
Queue . . . . . : QSYSOPR Program . . . . : *DSPMSG
Library . . . : QSYS Library . . . :
Severity . . . : 99 Delivery . . . : *HOLD
6. After the command swpprc *complete has completed, the switch to the remote site is
normally done. To check, use the command dspessdta on the local or remote partition.
666 IBM System Storage DS8000 Series: Copy Services in Open Environments
Verify that the Metro Mirror direction is reversed, and that the remote partition ITCHA3 is
currently the production node. This is shown in Figure B-43.
7. You can also check the status of IASP in both the local and the remote partition by using
the wrkcfgsts command, as described in “Checking the Toolkit setup” on page 644. Check
that the IASP at the remote site is available and that the IASP at the local site is not
available (varied off).
For more information about Global Mirror refer to Part 6, “Global Mirror” on page 241. For
more information about IASPs refer to “Independent Auxiliary Storage Pools (IASPs)” on
page 635.
For more information about System i5 clusters refer to “Clusters” on page 633.
Note that it is only necessary that IASP on both partitions (local and remote) reside on
external storage. Still, the sysbas for each partition can be on internal disks or on external
disk systems. For more information about sysbas refer to “Independent Auxiliary Storage
Pools (IASPs)” on page 635.
The solution is implemented with the System i Copy Services Toolkit available for purchase
from System i Client Technology Center in Rochester. For more information about the Toolkit
refer to “Solution description” on page 637 and “Implementation and usage” on page 641.
The solution provides continuous availability in case of planned and unplanned outages at the
local partition:
When a planned outage is scheduled for the local partition, the following actions are done
by the Toolkit:
a. Synchronize Global Copy on secondary volumes.
b. Vary on IASP (Global Mirror secondary volumes) at the remote partition.
c. Once IASP is available to the remote partition, start Global Copy to replicate the IASP
back to the local partition.
d. After the planned outage is finished, vary on the local IASP, which now contains the
updates that were made at the recovery site, at the local partition.
e. Start the replication again in the original direction.
For unplanned outages the toolkit performs the following actions:
a. Ensure that Global Mirror secondary volumes are consistent and as up-to-date as
possible by performing a fast reverse restore from FlashCopy target volumes to the
Global Mirror secondary volumes.
b. Vary on IASP (Global Mirror secondary volumes) at the remote partition where
production application continues to run using this IASP.
c. When the local site is back, replicate updated IASP to the local site using a Global
Copy.
d. Vary on IASP at the local partition where production can now resume.
668 IBM System Storage DS8000 Series: Copy Services in Open Environments
An overview of IASP Global Mirror is shown on Figure B-44.
GM
secondary
Global IASP
Mirror
IASP Flash
copy
Solution benefits
Global Mirror of an IASP includes the following benefits:
Since replication to the remote site is done asynchronously, the impact on application
response time is minimal.
For System i5 clients who require long-distance replication, Global Mirror, thanks to its
inherent asynchronous mode, is the most suitable solution amongst other long-distance
replication solution offerings.
Clients get the possibility to balance between RPO and bandwidth: If clients can cope with
some bigger RPO, they can use relatively narrow bandwidth for links between the two
sites. On the other hand, if they use large bandwidth, RPO will be very small and the loss
of data at recovery is almost negligible.
Basically, recovery takes as long as is needed to vary on an IASP. This recovery is
significantly shorter with Global Mirror than with solutions relying on Boot from SAN, or an
external loadsource, where IPL is needed for recovery.
In a production partition we run critical applications, while in a remote partition we can run
a non production workload like testing or development. This is different from
implementation using Boot from SAN where the recovery partition has to remain in
standby, typically without any workload.
Global Mirror requires a carefully performed sequence of steps during the recovery
procedure in unplanned outage situations. For clients with a huge number of volumes, it is
virtually impossible to perform these steps manually. The System i Copy Services Toolkit,
which is used as part of the solution, provides full automation of all the tasks needed to
recover the IASP.
Once this solution is established it requires very little maintenance and administration.
Therefore clients may prefer it to other high availability solutions available for System i5.
The Client Technology Center can perform an in-depth analysis of bandwidth requirements by
using testing workload PEX, if more accurate sizing is needed.
Other planning considerations and requirements for this solution are the same as for the
solution for Metro Mirror of an IASP described in “Planning and requirements” on page 639,
with the following change: for this solution a Global Mirror license is needed for the local and
remote DS systems, and FlashCopy license is needed for the remote DS system.
Note: For information about the toolkit contact the Client Technology Center via the
following Internet page:
http://www-03.ibm.com/servers/eserver/services/iseriesservices.html
Considerations
The same considerations as in “Considerations” on page 640 for Metro Mirror of an IASP also
apply for the Global Mirror of an IASP.
Important: When using Copy Services functions such as Metro Mirror, Global Mirror, or
FlashCopy for the replication of the load source unit or other i5/OS disk units within the
same DS6000/DS8000 or between two or more DS6000 and DS8000 systems, the source
volume and the target volume characteristics must be identical. The target and source
must be of matching capacities and matching protection types.
Also, once a volume is assigned to an i5/OS partition and added to that partition’s
configuration, its characteristics must not be changed. If there is a requirement to change
some characteristic of a configured volume, it must first be completely removed from the
i5/OS configuration. After the characteristic changes are made, for example, protection type,
capacity, and so on, by destroying and recreating the volume or by utilizing the DS CLI, the
volume can be reassigned to the i5/OS configuration. To simplify the configuration, we
recommend a symmetrical configuration between two IBM Storage systems, creating the
same volumes with the same volume ID that decides the LSS_ID.
670 IBM System Storage DS8000 Series: Copy Services in Open Environments
Implementation and usage
This solution is implemented as a service included with System i Copy Services Toolkit. The
installation steps for the Toolkit for Global Mirror are similar to those steps explained for Metro
Mirror in “Implementation and usage” on page 671.
After the Toolkit for Global Mirror is installed and tested you are able to perform the switch to
the remote site during planned and unplanned outages. The switch to remote site and the
failback to the production site are achieved with a series of actions performed by the toolkit.
They are started by toolkit commands swpprc and falovrgmir.
The Global Mirror status and toolkit setup status can be checked by issuing the commands
chkpprc and dspessdta, respectively.
Results of toolkit actions during the switch, or for checking the status after the switch, can be
observed in the Toolkit log by using a viewlog command.
The Toolkit triggers the different tasks by communicating with the System i5 HMC, performing
System i5 commands within the partition, and communicating with external disk systems by
using the DSCLI. This is illustrated in Figure B-7 on page 642.
Checking status
For a description on how to check the IASP status, refer to “Implementation and usage” on
page 641.
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
Parameter ENV required.
For more information about Metro Mirror refer to Part 4, “Metro Mirror” on page 107. For more
information about Boot from SAN refer to the IBM Redbook IBM System Storage DS6000
Series: Architecture and Implementation, SG24-6781.
Solution description
Because of the System i5 single level storage architecture, it is required that all disk units for
the local partition reside on external storage and that all of them are replicated (with Metro
Mirror) to the remote site.
672 IBM System Storage DS8000 Series: Copy Services in Open Environments
In the solution we review here, the loadsource is an external disk, is connected through Boot
from SAN IOP, and is also replicated to the remote site using Metro Mirror. We assume that
the standby partition in the System i5 (recovery partition) at the remote site has access to the
copy of the production volumes (Metro Mirror targets), and that Boot from SAN is done from
the Metro Mirror copy of the loadsource. This solution is depicted in Figure B-46.
Production Recovery
Metro
BfS
Mirror
BfS
In case of planned or unplanned outage at the production site, the stand-by partition at the
recovery site uses Boot from SAN to start the system from the Metro Mirror copy of
production volumes. After the boot is completed, the recovery system (stand-by partition) has
access to an exact clone of the production system, including database files, application
programs, user profiles, job queues, dataareas, and so on. Critical applications can continue
to run in this production system clone.
After planned outage is over or the unplanned outage is fixed, a Metro Mirror of all the disk
space is performed from the standby partition to the local partition. This copies back to
production system all the updates that occurred during the outage. The original local partition
can then be rebooted (Boot from SAN) from the updated primary volumes and will also get
access to primary volumes updated with the changes that accused during the outage.
For more information about System i5 single level storage refer to “Single-level storage” on
page 631. For more information about Boot from SAN refer to refer to the IBM Redbook IBM
System Storage DS6000 Series: Architecture and Implementation, SG24-6781.
Solution benefits
This solution offers the following major benefits:
It can be implemented without any updates or changes to the production i5/OS. If for any
reason you are not ready yet to set up your application in an IASP, you can still get the
benefits of a Metro Mirror solution by applying it for the entire partition disk space.
The solution does not require any special maintenance on the production or stand-by
system partition. Practically, the only required task is to set up the Metro Mirror relations
for all the volumes making up the partition entire disk space. Once done, no further actions
are required.
With the Metro Mirror technology and mechanisms, the performance impact on the
production system is usually smaller than with similar solutions by other external storage
For guidelines on how to size external storage for System i5, as well as for information about
how to collect System i5 performance data and which reports are needed for sizing external
storage, as well as requirements for BfS, refer to the IBM RedbookiSeries and IBM
TotalStorage: A Guide to Implementing External Disk on Eserver i5, SG24-7120.
Important: When using Copy Services functions such as Metro Mirror, Global Mirror, or
FlashCopy for the replication of the load source unit or other i5/OS disk units within the
same DS6000/DS8000 or between two or more DS6000 and DS8000 systems, the source
volume and the target volume characteristics must be identical. The target and source
must be of matching capacities and matching protection types.
Also, once a volume is assigned to an i5/OS partition and added to that partition’s
configuration, its characteristics must not be changed. If there is a requirement to change
some characteristic of a configured volume, it must first be completely removed from the
i5/OS configuration. After the characteristic changes are made, for example, protection type,
capacity, and so on, by destroying and recreating the volume or by utilizing the DS CLI, the
volume can be reassigned to the i5/OS configuration. To simplify the configuration, we
recommend a symmetrical configuration between two IBM Storage systems, creating the
same volumes with the same volume ID that decides the LSS_ID.
Considerations
Some other important considerations for this solution are as follows:
When doing a Metro Mirror of all the disk units in the local partition, not only the critical
application data but all i5/OS objects (including temporary objects, job queues, message
674 IBM System Storage DS8000 Series: Copy Services in Open Environments
queues, and so on) are synchronously copied to remote site. This usually impacts
production performance more than just doing a Metro Mirror of objects in an IASP. Some
installations can experience significant differences in performance between Metro Mirror
of entire disk space and Metro Mirror of an IASP.
Also, it should be noted that Metro Mirror of the entire disk space sometimes causes the
System i5 storage management to change the transfer sizes of IO operations. The transfer
size changes can have a significant performance impact to the production parititon.
Within the System i5 single level storage architecture, updates to application data are
performed in main memory. Updated database rows usually stay in main memory for
some time and are only written to disk as a result of a page swap. However, if a database
is journaled, changes to journaled objects are written immediately to the journal receiver
on disk.
When a switch to the remote site is performed at unplanned outages, the updates that
were still in main memory and not yet written to disk will not be reflected at the remote site,
since Metro Mirror only replicates what is on disk. However, with a database, since the
journaled objects are immediately written to disk and replicated by Metro Mirror, they will
be available at the remote site. Thus we strongly recommend journaling application
objects when using Metro Mirror. This applies whether you are mirroring the entire disk
space or just an IASP.
For more information about System i5 single level storage refer to “Single-level storage” on
page 631.
After a remote partition is IPLed from Metro Mirror secondary volumes, it has access to an
exact clone of the production partition, including i5/OS attributes and descriptions. Usually
some changes need to be done to the cloned partition for applications to run properly on
the recovery System i5. For more information about this topic, refer to the IBM
RedbookiSeries and IBM TotalStorage: A Guide to Implementing External Disk on
Eserver i5, SG24-7120.
Implementation
Since implementation includes several System i5 related tasks, we recommend that a System
i5 specialist be involved in implementing it. The implementation is done according to the
following steps:
1. For each System i5 production and remote partitions, install FC adapters and IOPs as
needed to implement Boot from SAN.
2. On the DS system define logical volumes for the System i5 and connect them to the
System i5.
3. Install i5/OS in the local partition or migrate an existing i5/OS to external disk.
4. Restore and start System i5 production applications.
5. Start Metro Mirror for all System i5 volumes. For more information about how to do this
refer to 12.1, “Basic Metro Mirror operation” on page 114. You can use DS CLI on System
i5 to perform this action.
For more information about the preceding topics refer to the IBM Redbooks iSeries and IBM
TotalStorage: A Guide to Implementing External Disk on Eserver i5, SG24-7120, and IBM
System Storage DS6000 Series: Architecture and Implementation, SG24-6781.
In this panel click the tab Tagged I/O and note the location of IOP or FC adapter for the
loadsource, as shown in Figure B-47.
On the Tagged I/O tab click the Select button at the tagged load source. This brings up the
panel Load Source Device. On the Load Source Device panel expand the unit, expand buses,
look for the slot that is tagged, select the slot, and click Properties. This is shown in
Figure B-48.
676 IBM System Storage DS8000 Series: Copy Services in Open Environments
This brings up the Physical I/O Properties panel, as shown in Figure B-49.
Click the I5/OS tab. You are presented with the description of IOP, as shown in Figure B-50.
Note the feature code 2847 that indicates Boot from SAN.
Note: The examples shown below were done on a DS8000, using LSS number 37. When
performing these actions on a DS6000, another LSS number will be used since DS6000
supports only LSS numbers 00 to 1F. Otherwise, the DS CLI scripts used for DS6000 are
the same as shown in these examples.
Note: A normal IPL in System i5 occurs after the system is powered down with the
power down System (PWRDWNSYS) command and no jobs have ended abnormally.
All other IPLs are abnormal for i5/OS. An abnormal IPL takes longer because of
additional recovery and verification needed.
2. By using the DS GUI or DS CLI, list the status of all the volumes in the local partition and
the recovery partition. The DS CLI command and a sample result are shown in
Figure B-51. Observe that both Metro Mirror primary and secondary volumes are in Full
Duplex state.
Note that n the example shown below we used the same scripts to perform actions on
primary and secondary DS systems. This can be done since the volumes on primary and
secondary DS have the same volume IDs. The action was performed on one or the other
DS by specifying relevant DS CLI profile as part of the DS CLI command. Each profile
contains IP addresses and image IDs of the local and remote DS storage systems.
Note: The DS CLI commands and scripts can be used from a System i5. To do this,
you need a separate partition in the System i5 (not the production partition or recovery
partition). The separate partition must have connectivity to both the local and remote
DS systems.
3. Suspend Metro Mirror for the local partition volumes. The DS CLI script for suspending the
volumes is shown in Figure B-52.
#
# suspend Metro Mirror
#
pausePPRC 3700-3707:3700-3707
Figure B-52 Script for suspend Metro mirror
678 IBM System Storage DS8000 Series: Copy Services in Open Environments
Performing the script and the results are shown in Figure B-53. Suspend Metro Mirror is
performed on the primary DS system.
4. Perform a Metro Mirror failover to the secondary DS system. For more information about
failover refer to 12.3, “Failover and Failback” on page 115. The DS CLI script for failover
and the results of running the script are shown in Figure B-54 and Figure B-55. Failover is
performed on a secondary DS system.
#
# Residency Mainz - failover Metor Mirror #
failoverPPRC -type mmir 3700-3707:3700-3707
Observe that both primary volumes (volumes belonging to the production partition) and
secondary volumes (belonging to the recovery partition) are in Source suspended state,
as shown in Figure B-56 and Figure B-57 on page 680.
Note that in our examples the DS primary system is 7503461 and the DS secondary
system is 7520781.
5. IPL the remote partition from the secondary volumes. To perform this, the secondary
volumes have to be connected to the standby partition. We recommend that the standby
partition contain two Boot form SAN (BfS) IOPs, each one of them connected to a copy of
a mirrored mate of load source. Make sure that one of the BfS IOPs is tagged as load
source in the System i5 HMC.
To IPL the remote partition perform the following steps from the System i5 HMC:
a. In the HMC navigation area expand Server and Partition and click Server
management to bring up the Server and Partition: Server Management panel. In this
panel expand Server, expand Partitions, and look for the remote partition icon.
b. Expand the view by clicking the plus sign next to the icon, then right-click the upper
icon and select Activate from the pull-down menu. This is shown on Figure B-58.
6. Once the remote partition is up and running, the applications can continue to run in the
recovery partition.
7. When the planned outage is finished, perform Metro Mirror failback to the primary volumes
at the production site. Doing this ensures that updates done at the remote partition during
the outage are now replicated back to the production partition. You may want to do a
Metro Mirror failback even before the outage is over, depending on the type of outage. The
DS CLI script used for the failback and performing the script are illustrated in Figure B-59
on page 681 and Figure B-60. Note that the failback is done on the secondary DS system.
680 IBM System Storage DS8000 Series: Copy Services in Open Environments
#
# Residency Mainz - failback MetrorMirror
#
failbackPPRC -type mmir 3700-3707:3700-3707
3. Fail over Metro Mirror to the primary volumes at the production site. The DS CLI script for
this action and results of its execution are shown in Figure B-54 on page 679 and
Figure B-62. This action is performed on the primary DS system.
5. IPL the production parititon and start the applications at the production site. Make sure
that the correct IOP for Boot from SAN is tagged for load source on the System i5 HMC,
as described earlier in this section.
Solution description
In this solution, the entire local partition disk space resides on external storage. The load
source is also external and connected to the System i5 via Boot from SAN (BfS). A Global
Mirror for all volumes belonging to the production partition is established with another DS
storage system located at the remote site.
682 IBM System Storage DS8000 Series: Copy Services in Open Environments
An important consideration here is the balance between the bandwidth to be provided for
Global Mirror, and RPO. It is essential to understand this issue and set the right expectations
when planning for this solution.
For more information about System i5 Boot from SAN refer to the IBM Redbook IBM System
Storage DS6000 Series: Architecture and Implementation, SG24-6781.
In planned or unplanned outage situations, a standby partition in the System i5 at the remote
site is IPLed from the Global Mirror secondary volumes. The Global Mirror copy of the load
source is also connected to the standby partition with Boot from SAN.
To ensure that the data to recover from are as up-to-date as possible, the solution includes
FlashCopy volumes of the Global Mirror targets. A fast restore from FlashCopy targets of the
Global Mirror secondary volumes is performed before IPL.
Production Recovery
GM secondary
Global
BfS
Mirror Flashcopy tgts
BfS
Solution benefits
Some of the benefits of establishing a Global Mirror for the entire disk space include:
The solution provides replication of production data over long distances while minimizing
production performance impact.
The solution offers the possibility to balance between RPO and bandwidth. If you can cope
with longer RPO, you can use relatively narrow bandwidth for links between the two sites.
On the other hand, if you have a large bandwidth, the RPO will be very small and of data
loss at recovery will be almost negligible.
Considerations about impact on System i5 production performances which apply to Metro
Mirror of entire disk space do not apply to Global Mirror of entire disk space. Therefore this
solution is recommended to System i5 customers. For more information about Metro
Mirror impact on System i5 performances refer to “Considerations” on page 674.
The solution does not require any special maintenance on the production or standby
partition. Practically, the only required task is to set up Global Mirror for the entire disk
space.
Since Global Mirror is entirely driven by the DS storage systems, this solution does not
use any processor or memory resources from the System i5 production and remote
For more information about the preceding topics you can refer to the IBM Redbook iSeries
and IBM TotalStorage: A Guide to Implementing External Disk on Eserver i5, SG24-7120.
Considerations
Other considerations for this solution are:
Recovery from Global Mirror secondary volumes in unplanned outage situation requires
additional steps compared to the recovery from Metro Mirror. The extra steps are required
to ensure a minimal data loss. In environments with large capacity and thus a large
number of volumes, it is extremely difficult to perform these steps by only issuing DS CLI
commands, and the use of an automation tool is recommended.
We also recommend using i5/OS journaling with this solution. For more information about
journaling, refer to “Considerations” on page 674.
Important: When using Copy Services functions such as Metro Mirror, Global Mirror, or
FlashCopy for the replication of the load source unit or other i5/OS disk units within the
same DS6000/DS8000 or between two or more DS6000 and DS8000 systems, the source
volume and the target volume characteristics must be identical. The target and source
must be of matching capacities and matching protection types.
Also, once a volume is assigned to an i5/OS partition and added to that partition’s
configuration, its characteristics must not be changed. If there is a requirement to change
some characteristic of a configured volume, it must first be completely removed from the
i5/OS configuration. After the characteristic changes are made, for example, protection type,
capacity, and so on, by destroying and recreating the volume or by utilizing the DS CLI, the
volume can be reassigned to the i5/OS configuration. To simplify the configuration, we
recommend a symmetrical configuration between two IBM Storage systems, creating the
same volumes with the same volume ID that decides the LSS_ID.
684 IBM System Storage DS8000 Series: Copy Services in Open Environments
Implementation and usage
Customers who want to implement Global Mirror for the entire System i disk space may use
the System i Copy Services Toolkit available for purchase from the System i Client
Technology Center in Rochester. The toolkit is described in “Global Mirror for an IASP” on
page 667. Once the Toolkit is installed, you can do a switchover to the remote site during
planned and unplanned outages by using the Toolkit commands.
Still, for smaller installations with only a few volumes, it is feasible to perform the switchover to
the remote site by simply using the DS CLI commands. We describe the steps to perform for
both situations: either using Toolkit or using DS CLI.
For information about how to set up Global Mirror refer to Part 6, “Global Mirror” on page 241.
If customers want all the steps listed to be performed automatically by one toolkit command,
they can request this as an additional service with the toolkit services.
Using DS CLI
Note: The examples shown below were done on a DS8000, using LSS numbers 37 and
39. When performing these actions on a DS6000, another LSS number will be used since
the DS6000 supports only LSS numbers 00 to 1F. Otherwise, DS CLI scripts used on the
DS6000 are the same as shown in these examples.
5. In planned outage situations, power down the local partition and suspend Global Mirror
before the failover. This eliminates the need to perform a fast restore of the FlashCopy.
6. IPL the remote partition from the Global Mirror secondary volumes using Boot from SAN.
For instruction on how to do this refer to “Switch to remote site at planned outages” on
page 678. Once the partition is IPLed, it will be a clone of the local partition, and
production work can continue in the cloned partition at the remote site.
686 IBM System Storage DS8000 Series: Copy Services in Open Environments
Using DS CLI
To use this:
1. On the remote DS storage system, perform a failback, as shown in Figure B-68.
If customers want all the steps listed to be performed automatically by one toolkit command,
they can request this as an additional service with the Toolkit services.
688 IBM System Storage DS8000 Series: Copy Services in Open Environments
3. List FlashCopy volumes to see the status of sequences and revertible volumes. The DS
CLI command and partial output are shown in Figure B-73.
Examine the sequence numbers and volumes revertible status, and consequently decide
to revert FlashCopy, commit FlashCopy, or not take any action. For more information refer
to Chapter 21, “Global Mirror options and configuration” on page 263. In our example all
sequence numbers are equal and we do not have revertible volumes, so we do not
perform any action here.
4. Reverse FlashCopy on remote DS, as shown in Figure B-74
5. IPL the remote partition from the Global Mirror secondary volumes by using BfS. Once IPL
is done, the production application can continue to run in the clone of the local partition.
FlashCopy of IASP
The downtime needed to save application data to tape is critical for many System i5
installations because saving to tape is usually a part of typical batch jobs running in those
environments. For many clients, this downtime window is generally very limited.
The solution described in this section is based on FlashCopy and ensures minimal downtime
for saving application data to tape.
For more information about FlashCopy, refer to Part 3, “FlashCopy” on page 35. For more
information about IASP refer to “Independent Auxiliary Storage Pools (IASPs)” on page 635.
Solution description
The setup for this solution consists of two System i5 partitions. Usually both partitions are
defined in the same physical System i5, but they could also reside in two separate units.
To take a backup of the application using the backup partition, follow these steps:
1. Perform a FlashCopy of the IASP.
2. Vary on the IASP FlashCopy targets in the backup partition.
3. Use the backup partition to save the data to tape without impacting the production
partition.
During the backup operation, the production application continues to run in production
partition.
We recommended varying off the IASP in the production partition just before taking the
FlashCopy. Doing so will minimize the time needed to bring up the application in the backup
partition.
Production IASP
Flash
copy
Backup IASP
Backup
Save
Note: For information about the Toolkit contact the Client Technology Center via the
following Internet page:
http://www.ibm.com/servers/eserver/services/iseriesservices.html
690 IBM System Storage DS8000 Series: Copy Services in Open Environments
Solution benefits
The most important benefits are as follows:
The production application downtime is only is as long as it takes to vary off the IASP,
perform a FlashCopy of the volumes in the IASP, and vary on IASP. Usually this time is
estimated to be about 30 minutes. This is very little compared to the downtime normally
experienced when saving to tape without a Save While Active function.
For more information about Save While Active refer to the iSeries Information Center on
the following Web page:
http://publib.boulder.ibm.com/iseries
The performance impact on the production application during the save to tape operation is
minimal since it is only influenced by the FlashCopy activity, which is mostly confined
within the DS system. Furthermore, this FlashCopy activity is typically very low since save
to tape operations are typically performed at night or during light workload periods.
This solution can be implemented together with Backup, Recovery and Media Services for
iSeries (BRMS), a System i5 software for saving application data to tape.
For more information about BRMS refer to the iSeries Information Center at the following
web page:
http://publib.boulder.ibm.com/iseries/
Considerations
This solution requires you to vary off the production IASP every time a FlashCopy of the IASP
is taken. Some clients prefer not to vary off the IASP before FlashCopy, and therefore may
experience a longer time to vary on the IASP in the backup partition.
Other considerations listed under “Considerations” on page 640 for Metro Mirror of IASP also
apply here.
Important: When using Copy Services functions such as Metro Mirror, Global Mirror, or
FlashCopy for the replication of the load source unit or other i5/OS disk units within the
same DS6000/DS8000 or between two or more DS6000 and DS8000 systems, the source
volume and the target volume characteristics must be identical. The target and source
must be of matching capacities and matching protection types.
The following should be taken into account when planning and implementing such combined
solutions:
System i Copy Services Toolkit for both Metro Mirror (or Global Mirror) and FlashCopy
should be purchased.
If FlashCopy is implemented at the remote site, the solution typically needs two remote
System i5 partitions: one for recovery from Metro Mirror (or Global Mirror) targets and one
for FlashCopy. Each of them requires dedicated FC adapters.
Regardless of whether FlashCopy is implemented at the local or remote site, it requires
that the production IASP be varied off before taking the FlashCopy. For more information
refer to “Considerations” on page 691.
The toolkit installation steps are described in “Implementation and usage” on page 641.
Once the toolkit is installed and tested, the following actions are performed to check the
solution components and perform a save to tape in the backup partition.
Refer to “Checking the solution components” on page 642. To check the FlashCopy status
use the command dspessdta.
692 IBM System Storage DS8000 Series: Copy Services in Open Environments
In a Telnet session with the System i5 partition, enter the command prpiaspbck. You are
presented with a screen where you can insert the IASP name, as shown in Figure B-76.
Press Enter to start executing the command.
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
Parameter IASPNAME required.
While executing the prpiaspbck command, the toolkit performs the following:
Vary off the production IASP.
Make a FlashCopy of the volumes in the DS system by running a DS CLI script in the
backup partition.
Add IOPs and FC adapters to the backup partition that will attach the FlashCopy of the
production IASP volumes. For more information about IOPs refer to “Input Output
Processors” on page 632.
Vary on the production IASP.
Vary on the FlashCopy of the production IASP in the backup partition.
After completion of the prpiaspbck command, check the IASP status in the backup partition.
For information about how to check this refer to “Checking the solution components” on
page 642.
Saving to tape
At this stage the backup partition is ready for taking a backup of the production application
based on the copy (FlashCopy) of the production IASP.
The actual backup can be done with BRMS or by using i5/OS commands.
For more information about System i5 backups refer to iSeries Information Center at the
following Web page:
http://publib.boulder.ibm.com/iseries
You can also consult the manual IBM Systems - iSeries Backup and Recovery Version 5
Revision 4, SC41-5304-08.
For more information about BRMS refer to the iSeries Information Center and the manual IBM
Systems - iSeries Backup Recovery and Media Services for iSeries Version 5,
SC41-5345-05.
Backups taken in the backup partition can be restored to the production partition. Single
System i5 objects from these backups can be restored to production partition, if needed.
In a Telnet session established with the System i5 partition, enter the command rsmiaspbck.
You are presented with a screen where you can insert the IASP name, as shown in
Figure B-77. Press Enter to start executing the command.
1. User tasks
2. Office tasks
3. General system tasks
4. Files, libraries, and folders
5. Programming
6. Communications
7. Define or change the system
8. Problem handling
9. Display a menu
10. Information Assistant options
11. iSeries Access tasks
Selection or command
===> rsmiaspbck
While the rsmiaspbck command execution takes place, the following actions are performed
by the toolkit:
Vary off the FlashCopy of the production IASP, which is attached to the backup node.
Remove the FlashCopy for the volumes in the DS unit by running the dscli rmflash
script.
IPL the backup node.
Once rsmiaspbck has completed its execution, the backup partition is in the appropriate state
for the next backup through the prpiaspbck command.
694 IBM System Storage DS8000 Series: Copy Services in Open Environments
FlashCopy of the entire disk space
In this solution, FlashCopy is implemented for the entire disk space attached to a System i5
parititon. This solution will minimize the production partition downtime for taking backups. For
more information about FlashCopy refer to Chapter 5, “FlashCopy overview” on page 37.
Solution description
The setup for this solution consists of two System i5 partitions. Usually both partitions reside
on the same physical System i5, but they could be also on different units. We call the
production partition the partition where the production applications run. The other partition is
called the backup parititon. The backup partition is standby and typically does not run any
workload. All disks attached to the production partition reside on the external DS storage
system and Boot from SAN is implemented.
The standby backup partition also contains Boot from SAN IOP.
For more information about Boot from SAN refer to the IBM Redbook IBM System Storage
DS6000 Series: Architecture and Implementation, SG24-6781.
After being IPLed, the backup partition is a clone of the production partition. Any backup
taken in the backup parititon will reflect actual and current production data, and later it will be
possible to restore this backup to the production partition.
For more information about the previous considerations, refer to the IBM Redbook iSeries and
IBM TotalStorage: A Guide to Implementing External Disk on Eserver i5, SG24-7120.
Production
BfS
Flash
copy
Backup
Backup
BfS
Save
Solution benefits
This solution offers the following major benefits:
The production application downtime is only is as long as it takes to power down the
production partition, take a FlashCopy of the production volumes, and IPL the production
partition (IPL is normal). Usually, this time is much shorter than the downtime experienced
when saving to tape without a Save While Active function.
For more information about Save While Active refer to the iSeries Information Center on
the following Web page:
http://publib.boulder.ibm.com/iseries/
The performance impact on the production application during the save to tape operation is
minimal since it is only influenced by the FlashCopy activity, which is mostly confined
within the DS system. Furthermore, this FlashCopy activity is typically very low since save
to tape operations are normally performed at night or during light workload periods.
This solution can be implemented together with Backup, Recovery, and Media Services for
iSeries (BRMS), a System i5 software for saving application data to tape.
For more information about BRMS refer to the iSeries Information Center at the following
Web page:
http://publib.boulder.ibm.com/iseries/
696 IBM System Storage DS8000 Series: Copy Services in Open Environments
Planning and requirements
It is important to adequately size the primary and secondary DS systems. Also, since Boot
from SAN is used in this solution, hardware and software requirements for Boot form SAN
need to be taken into consideration for both the production and backup partitions.
For guidelines on how to size external storage for System i5 as well as more information
about requirements for BfS refer to the IBM Redbook iSeries and IBM TotalStorage: A Guide
to Implementing External Disk on Eserver i5, SG24-7120.
Considerations
This solution requires powering down the production partition each time a FlashCopy is
taken, and then IPLing the partition after completion of the FlashCopy. This procedure is
strongly recommended in order to ensure a correct IPL of the backup partition later in the
process.
Some installations cannot tolerate a daily IPL of their i5/OS production system, and
consequently cannot stick to the recommended procedure. In this case we suggest that they
implement IASP and use the solution described in “FlashCopy of IASP” on page 689.
Important: When using Copy Services functions such as Metro Mirror, Global Mirror, or
FlashCopy for the replication of the load source unit or other i5/OS disk units within the
same DS6000/DS8000 or between two or more DS6000 and DS8000 systems, the source
volume and the target volume characteristics must be identical. The target and source
must be of matching capacities and matching protection types.
Also, once a volume is assigned to an i5/OS partition and added to that partition’s
configuration, its characteristics must not be changed. If there is a requirement to change
some characteristic of a configured volume, it must first be completely removed from the
i5/OS configuration. After the characteristic changes are made, for example, protection type,
capacity, and so on, by destroying and recreating the volume or by utilizing the DS CLI, the
volume can be reassigned to the i5/OS configuration. To simplify the configuration, we
recommend a symmetrical configuration between two IBM Storage systems, creating the
same volumes with the same volume ID that decides the LSS_ID.
Implementation
Since the implementation of this solution implies several System i5 related tasks, we
recommend involving a System i5 specialist.
Note that if BRMS is used for saving to tape from the backup partition, some additional steps
are required to ensure that backups will correctly restore in the production partition. These
steps are described in the IBM Redbook iSeries and IBM TotalStorage: A Guide to
Implementing External Disk on Eserver i5, SG24-7120.
Note: The examples shown here were done on a DS8000, using LSS numbers 37 and 39.
When performing these actions on a DS6000, another LSS number will be used since the
DS6000 supports only LSS numbers 00 to 1F. However, DS CLI scripts used on the
DS6000 are the same as those shown in these examples.
698 IBM System Storage DS8000 Series: Copy Services in Open Environments
CMUC00137I mkflash: FlashCopy pair 3705:3905 successfully created.
CMUC00137I mkflash: FlashCopy pair 3706:3906 successfully created.
CMUC00137I mkflash: FlashCopy pair 3707:3907 successfully created.
3. After the FlashCopy is performed, you may want to list the FlashCopy relationship of the
production volumes. The DS CLI script to perform this and the command to invoke the
script in Windows, as well as the output of the command, are shown in Example B-3 and
Example B-3.
4. IPL the production partition and start the production applications. For instructions on how
to do this refer to “Switch to remote site at planned outages” on page 678.
5. Next IPL the backup partition from the FlashCopy target volumes, via BfS. Once the
partition is IPLed, save the entire system or specific objects to tape using BRMS or native
i5/OS commands.
For more information about System i5 backups, refer to the iSeries Information Center on
the following Web page:
http://publib.boulder.ibm.com/iseries
You can also consult the IBM Systems - iSeries Backup and Recovery Version 5
Revision 4, SC41-5304-08.
700 IBM System Storage DS8000 Series: Copy Services in Open Environments
C
The DS HMC can be set up to send SNMP traps to up to two defined IP addresses. eRCMF
(see Chapter 36, “IBM TotalStorage Rapid Data Recovery for UNIX and Windows” on
page 575) is listening to SNMP traps of the DS8000. Also, a Network Management program,
like Tivoli NetView®, can be used to catch and process the SNMP traps.
If one or several links of multiple links are interrupted, a trap 100, as shown in Example C-1, is
posted and will indicate that the redundancy is degraded. The RC column in the trap will
represent the reason code for the interruption of the link. The reason codes are listed in
Table C-1 on page 704.
If all links are interrupted, a trap 101, as shown in Example C-2, is posted. This event is
indicating that no communication between primary and secondary system is possible any
more.
702 IBM System Storage DS8000 Series: Copy Services in Open Environments
Once the DS8000 can communicate again via any of the links, trap 102, as shown in
Example C-3, is sent once the interrupted links are available again.
00 No path.
02 Initialization failed. ESCON link reject threshold exceeded when attempting to send ELP or RID
frames.
07 Secondary CU SS ID mismatch or failure of the I/O that collects secondary information for validation.
08 ESCON link is offline. This is caused by the lack of light detection coming from a host, peer, or switch.
0A The primary control unit port or link cannot be converted to channel mode since a logical path is
already established on the port or link. The establish paths operation will not be retried within the
control unit automatically.
11 Reserved.
12 Reserved.
16 Fibre Channel Path Secondary Adapter not PPRC capable. This could be due to:
Secondary Adapter not configured properly, or does not have the correct microcode loaded.
The secondary adapter is already a target of 32 different ESS, DS8000, and DS6000.
704 IBM System Storage DS8000 Series: Copy Services in Open Environments
Remote copy events
If you have configured Consistency Groups and a volume within this Consistency Group is
suspended due to a write error to the secondary device, trap 200, as shown in Example C-4,
will be sent out. One trap per LSS is configured with the Consistency Group option. This trap
could be handled by automation software like eRCMF to freeze this Consistency Group.
Trap 202, as shown in Example C-5, will be sent out if a remote Copy Pair goes into a
suspend state. The trap contains the serial number (SerialNm) of the primary and secondary
machine, the LSS (LS), and the logical device (LD). To avoid SNMP trap flooding, the number
of SNMP traps for the LSS will be throttled. The complete suspended pair information is
represented in the summary. The last row of the trap represents the suspend state for all pairs
in the reporting LSS. The suspended pair information contains a hexadecimal string of a
length of 64 characters. By converting this hex string into binary, each bit will represent a
single device. If the bit is 1, then the device is suspended; otherwise the device is still in full
duplex mode.
Example: C-5 Trap 202: Primary PPRC devices on LSS suspended due to error
Primary PPRC Devices on LSS Suspended Due to Error
UNIT: Mnf Type-Mod SerialNm LS LD SR
PRI: IBM 2107-922 75-20781 11 00 03
SEC: IBM 2107-9A2 75-ABTV1 21 00
Start: 2005/11/14 09:48:05 CST
PRI Dev Flags (1 bit/Dev, 1=Suspended):
C000000000000000000000000000000000000000000000000000000000000000
Example: C-6 Trap 210: Asynchronous PPRC initial Consistency Group successfully formed
2005/11/14 15:30:55 CET
Asynchronous PPRC Initial Consistency Group Successfully Formed
UNIT: Mnf Type-Mod SerialNm
IBM 2107-922 75-20781
Session ID: 4002
Trap 211, as shown in Example C-7, will be sent out if the Global Mirror setup got into an
severe error state, where no attempts will be done to form a Consistency Group.
Example: C-8 Trap 212: Asynchronous PPRC Consistency Group failure - Retry will be attempted
Asynchronous PPRC Consistency Group Failure - Retry will be attempted
UNIT: Mnf Type-Mod SerialNm
IBM 2107-922 75-20781
Session ID: 4002
Trap 213, as shown in Example C-9, will be sent out when a Consistency Group in a Global
Mirror environment could be formed after a previous Consistency Group formation failure.
Example: C-9 Trap 213: Asynchronous PPRC Consistency Group successful recovery
Asynchronous PPRC Consistency Group Successful Recovery
UNIT: Mnf Type-Mod SerialNm
IBM 2107-9A2 75-ABTV1
Session ID: 4002
Trap 214, as shown in Example C-10, will be sent out if a Global Mirror copy session is
terminated using the command rmgmir or the corresponding GUI function.
Trap 215, as shown in Example C-11, will be sent out if in the Global Mirror environment the
master has detected a failure to complete the FlashCopy commit. The trap will be sent out
after a number of commit retries have failed.
Example: C-11 Trap 215: Asynchronous PPRC FlashCopy at remote site unsuccessful
Asynchronous PPRC FlashCopy at Remote Site Unsuccessful
A UNIT: Mnf Type-Mod SerialNm
IBM 2107-9A2 75-ABTV1
Session ID: 4002
Trap 216, as shown in Example C-12, will be sent out if a Global Mirror master cannot
terminate the Global Copy relationship at one of his subordinates (slaves). This might occur if
the master is terminated with rmgmir but the master cannot terminate the copy relationship on
the subordinate. The user may need to run a rmgmir against the subordinate to prevent any
interference with other Global Mirror sessions.
706 IBM System Storage DS8000 Series: Copy Services in Open Environments
Trap 217, as shown in Example C-13, will be sent out if a Global Mirror copy environment was
suspended by the DS CLI command pausegmir or the corresponding GUI function.
Appendix D. Licensing
In this appendix we describe how the licensing functions for Copy Services for the DS8000
Series are arranged.
Licensed functions require the selection of a DS8000 series feature number (IBM 2107) and
the acquisition of DS8000 Series Function Authorization (IBM 2244) feature numbers:
The 2107 licensed function indicator feature number enables technical activation of the
function subject to the client applying an activation code made available by IBM.
The 2244 function authorization feature numbers establish the extent of IBM authorization
for that function on the 2107 machine for which it was acquired.
Table D-1 lists the DS8000 series feature numbers and corresponding DS8000 Series
Function Authorization feature numbers.
Attention: For the DS8000 Turbo models 931, 932, and 9B2, the Metro Mirror and Global
Mirror features are now licensed separetely as indicated in Table D-1. For the former 92X
and 9A2 models, support for Metro Mirror and Global Mirror continues to be provided with
2244 model RMC features 740x and 7410, and indicator feature 0740.
The following is the breakdown of DS8000 Series Function Authorization feature numbers:
OEL: Operating Environment License
– OEL - inactive 7020
– OEL - 1 TB 7021
– OEL - 5 TB 7022
– OEL - 10 TB 7023
– OEL - 25 TB 7024
– OEL - 50 TB 7025
– OEL - 100 TB 7030
FICON: z/OS attachment
– FICON Attachment 7090
710 IBM System Storage DS8000 Series: Copy Services in Open Environments
PAV: Parallel Access Volumes
– PAV - inactive 7800
– PAV - 1 TB 7801
– PAV - 5 TB 7802
– PAV - 10 TB 7803
– PAV - 25 TB 7804
– PAV - 50 TB 7805
– PAV - 100 TB 7810
712 IBM System Storage DS8000 Series: Copy Services in Open Environments
Authorized level
All Copy Services functions require licensing to be activated. This means that the customer
must purchase a license for the appropriate level of storage for each Copy Service function
that is required. They then have to install the License Key generated using the Disk Storage
Feature Activation application (DSFA). Another consideration relates to the authorized level
required. In most cases the total capacity installed must be licensed. This is the total capacity
in decimal TB equal to or greater than the actual capacity installed, including all RAID parity
disks and hot spares.
An exception may be where a mix of both System z and open systems hosts are using the
same storage server. In this case it is possible to acquire Copy Services licenses for just the
capacity formatted for CKD, or just the capacity formatted for FB storage. This implies that the
licensed Copy Services function is needed only for open systems hosts, or only for System z
hosts. If, however, a Copy Services function is required for both CKD and FB, then that Copy
Services license needs to match the total configured capacity of the machine. The
authorization level is maintained by the licensed code in the controller and the DSFA
application.
For example, the actual capacity is 15 TB, used for both CKD and FB, the scope for the OEL
is therefore a type of ALL and the installed OEL must be at least 15 TB. If the client has split
storage allocation, with 8 TB for CKD, and only CKD storage is using FlashCopy, then the
scope type for the PTC license can be set to CKD. Now the PTC license can be purchased at
the CKD level of 8 TB. However, this means that no open systems hosts can use the
FlashCopy function.
The actual ordered level of any Copy Service license can be any level above that required or
installed. Licenses can be added and have their capacities increased, non-disruptively to an
installed system.
Important: A decrease in the scope or the capacity of a license requires a disruptive IML
of the DS8000 Storage Facility Image.
Charging example
A client can choose to purchase any or all DS8000 licenses at an authorization level at or
later than the installed raw disk capacity. It may be more cost effective to pre-install
authorization to use greater than the currently installed storage capacity.
A simple rule for charges is remembering that the required authorization level depends upon
the license scope:
If the license scope is FB, the authorization level must be equal to or greater than the total
amount of physical capacity within the system that will be logically configured as FB.
If the license scope is CKD, the authorization level must be equal to or greater than the
total amount of physical capacity within the system that will be logically configured as
CKD.
If the license scope is ALL, the authorization level must be equal to or greater than the
total system physical capacity.
The dual Storage Facility Image (SFI) version of DS8300 (the model 9Bx) uses the same
logic, in this case added between the two SFIs. So all of the above apply to each SFI
separately. It is possible to license Copy Services for a single scope in one SFI.
714 IBM System Storage DS8000 Series: Copy Services in Open Environments
E
The Advanced Copy Services functions that are available on the ESS 800 are also available
on the DS8000. Although the functions are still available, there are also some differences that
need to be considered in replacing your ESS CLI with a DS CLI. These are:
Point-in-Time Copy (FlashCopy) does not support Consistency Groups on the GUI.
Fibre Channel is used for Metro Mirror, Global Mirror, and Metro/Global Copy.
The GUI runs real-time only (tasks cannot be saved), while the CLI can be invoked using a
saved script.
The DS CLI supports both the ESS 800/750 and the DS8000.
ESS 800 must be at Licensed Internal Code (LIC) level 2.4.3.15 or later.
You can use ESS CLI to display the contents of each saved task and write the contents to a
file. See Example E-2.
716 IBM System Storage DS8000 Series: Copy Services in Open Environments
You can also check your saved tasks via the ESS GUI. See Figure E-1.
Highlight the task and click the information panel. See Figure E-2.
Also review the specific server scripts (depending on the OS) that perform tasks that set up
and execute saved ESS CLI saved tasks. You may need to edit or translate these scripts in
order to run your DS CLI saved tasks.
Important: On the ESS 800, open systems volume IDs are given in an 8-digit format,
xxx-sssss, where xxx is the LUN ID and sssss is the serial number of the ESS 800. In the
example used in this appendix the volumes shown are 004-23953 to 005-23953. These
volumes are open systems or fixed block volumes. When referring to them in the DS CLI,
you must add 1000 to the volume ID, so volume 004-23953 is volume ID 1004 and volume
005-23953 is volume ID 1005. This is very important because on the ESS 800, the
following address ranges are actually used:
0000 to 0FFF System z CKD volumes (4096 possible addresses)
1000 to 1FFF Open systems fixed block LUNs (4096 possible addresses)
If we intend to use FlashCopy to copy ESS LUN 004-23953 onto 005-23953 using the DS
CLI, we must specify 1004 and 1005. If instead we specify 0004 and 0005, we will actually
run the FlashCopy against CKD volumes. This may result in an unplanned outage on the
System z host that was using CKD volume 0005.
The ESS CLI command, show task, will show the correct value for the volume ID.
Note: The DS GUI (via the DS HMC) does not save Copy Services tasks like the ESS GUI.
You can only use the DS GUI for Copy Services in real-time mode.
Also translate (if needed) and edit the individual server scripts that will set up and execute the
saved DS CLI scripts. See Table E-1 for an example of command translation.
DS CLI commands
Example E-3 is the translation to DS CLI commands.
718 IBM System Storage DS8000 Series: Copy Services in Open Environments
ESS/DS CLI comparison
Table E-2 shows a brief comparison of the major components between the ESS CLI and the
DS CLI.
list featurecode lsda, lshba, lsioencl, lsuser, The 2107 CLI commands can
mkuser, rmuser, chuser, display feature codes when the
lsstgencl appropriate parameters are
used with the commands.
list task NA
show task NA
rsTestConnection lsaavailpprcport
720 IBM System Storage DS8000 Series: Copy Services in Open Environments
Note 1: Volume space configuration is a primary difference between 2105 and 2107. Like
the 2105, a 2107 Storage Facility Image volume space contains a RAID 5 or RAID 10 Array
and Rank that are configured from an Array Site. For the 2105, one command configures
an Array Site into a RAID array and Rank. For the 2107, one command configures an Array
Site into an Array, and a second command configures an Array into a Rank. For 2105, a
Rank is configured as fixed block or CKD. For the 2107, a Rank is assigned to a
user-defined Extent Pool object, which the user defines as either the fixed block or CKD
storage type. The interleave volume construct does not exist for 2107. For the 2105, a
volume is configured from a specific Rank, and cannot span Rank boundaries. For the
2107, a volume is configured from an Extent Pool. An Extent Pool can contain multiple
Ranks. A 2107 volume consists of one or more Extents that can be allocated from one or
more Ranks. A fixed block Extent is 1 GB (128 logical blocks). Each block contains 512
bytes of usable data space.
For 2105, a Rank is either assigned to server 0 or server 1, depending on the Array Site
location. A 2105 Rank is assigned to one of 32 possible LSS IDs, depending on the device
adapter pair location and storage type configuration.
For 2107, an Extent Pool is assigned to server 0 or server 1. A Rank that is configured from
any Array Site can be assigned to a server 0 or 1 Extent Pool. Array Site position and
device adapter pairs are not factors for the Rank to Extent Pool assignment. A volume that
is created from a server 0 Extent Pool is assigned to an even-numbered LSS ID. A volume
created from a server 1 Extent Pool is assigned to an odd numbered LSS ID. A user must
define at least two Extent Pools (0 and 1), but can define as many Extent Pools as there
are Ranks. For 2105, a user can delete a Rank but cannot delete a volume. For 2107, a
user can delete a single volume, Rank, or Extent Pool. The 2107 CLI showrank and
showextpool commands include a metrics parameter that returns the performance counter
values for a specified Rank or Extent Pool ID. The metrics parameter provides the means
to monitor Rank and Extent Pool performance statistics.
Note 2: Like 2105, a 2107 Fibre Channel SCSI-FCP I/O port can be configured for either
the point-to-point/switched fabric or FC-AL connection topologies. A port that uses the
point-to-point/switched fabric topology can be simultaneously used for OS host system I/O
and for PPRC path configurations. The Fibre Channel SCSI-FCP IO port only allows
identified host system ports to access volumes. A host system port WWPN must be
identified (registered) to each ESS IO port through which volume access is intended. Host
system port WWPN identification is accomplished by the CLI mkhostconnect command.
Note 3: A 2107 storage facility image can contain up to 65,536 volumes. A 2105 Storage
Unit can contain up to 4096 FB volumes and 4096 CKD volumes. Otherwise, the 2105 and
2107 volume definitions and characteristics are essentially identical.
The 2107 CLI provides a specific set of volume commands for each storage type, fixed
block or CKD, as a means to clarify input parameter and output device adapter definitions.
The 2107 CLI showfbvol and showckdvol commands include a metrics parameter that
returns the performance counter values for a specified volume ID. The metrics parameter
provides the means to monitor volume performance statistics.
The volume group construct also exists in the 2105 internal code, but the construct is not
externalized by the 2105 Specialist or CLI commands.
For 2107 fixed block volumes, a Volume Group must be configured as either SCSI-mask or
SCSI-map-256, depending whether the volume group is accessed by a SCSI-FCP host
port that used the report LUNs or poll LUNs access method protocol.
722 IBM System Storage DS8000 Series: Copy Services in Open Environments
Related publications
The publications listed in this section are considered particularly suitable for a more detailed
discussion of the topics covered in this book.
IBM Redbooks
For information about ordering these publications, see “How to get IBM Redbooks” on
page 725. Note that some of the documents referenced here may be available in softcopy
only.
The IBM TotalStorage DS8000 Series: Implementation, SG24-6786
IBM TotalStorage Business Continuity Solutions Guide, SG24-6547
IBM System Storage Solutions Handbook, SG24-5250
The IBM TotalStorage DS8000 Series: Copy Services with IBM Eserver zSeries,
SG24-6787
If you are implementing Copy Services in a mixed technology environment you may be
interested in referring to the following manuals on the ESS and DS6000.
IBM TotalStorage Enterprise Storage Server Implementing ESS Copy Services in Open
Environments, SG24-5757
IBM TotalStorage Enterprise Storage Server Implementing ESS Copy Services with
IBM Eserver zSeries, SG24-5680
The IBM TotalStorage DS6000 Series: Copy Services with IBM Eserver zSeries,
SG24-6782
The IBM TotalStorage DS6000 Series: Copy Services in Open Environments Redbook,
SG24-6783
DFSMShsm ABARS and Mainstar Solutions, SG24-5089
Practical Guide for SAN with pSeries, SG24-6050
Fault Tolerant Storage Multipathing and Clustering Solutions for Open Systems for the IBM
ESS, SG24-6295
Implementing Linux with IBM Disk Storage, SG24-6261
Linux with zSeries and ESS: Essentials, SG24-7025
Other publications
These publications are also relevant as further information sources. Note that some of the
documents referenced here may be available in softcopy only.
IBM System Storage DS8000 Command-Line Interface User´s Guide, SC26-7916
IBM System Storage DS8000: Host Systems Attachment Guide, SC26-7917
IBM System Storage DS8000: Introduction and Planning Guide, GC35-0515
IBM System Storage Multipath Subsystem Device Driver User’s Guide, SC30-4131
IBM System Storage DS8000: User’s Guide, SC26-7915
Online resources
These Web sites and URLs are also relevant as further information sources:
IBM Disk Storage Feature Activation (DSFA) Web site:
http://www.ibm.com/storage/dsfa
The PSP information:
http://www-1.ibm.com/servers/resourcelink/svc03100.nsf?OpenDatabase
Documentation for the DS8000:
http://www.ibm.com/servers/storage/support/disk/2107.html
The Interoperability Matrix:
http://www.ibm.com/servers/storage/disk/ds8000/interop.html
Fibre Channel host bus adapter firmware and driver level matrix:
http://knowledge.storage.ibm.com/servers/storage/support/hbasearch/interop/hbaSearch.do
ATTO:
http://www.attotech.com/
Emulex:
http://www.emulex.com/ts/dds.html
JNI:
http://www.jni.com/OEM/oem.cfm?ID=4
QLogic:
http://www.qlogic.com/support/ibm_page.html
IBM:
http://www.ibm.com/storage/ibmsan/products/sanfabric.html
McDATA:
http://www.mcdata.com/ibm/
Cisco:
http://www.cisco.com/go/ibm/storage
CIENA:
http://www.ciena.com/products/transport/shorthaul/cn2000/index.asp
Nortel:
http://www.nortelnetworks.com/
ADVA:
http://www.advaoptical.com/
724 IBM System Storage DS8000 Series: Copy Services in Open Environments
How to get IBM Redbooks
You can search for, view, or download Redbooks, Redpapers, Hints and Tips, draft
publications and Additional materials, as well as order hardcopy Redbooks or CD-ROMs, at
this Web site:
ibm.com/redbooks
728 IBM System Storage DS8000 Series: Copy Services in Open Environments
Enterprise level 582 Fibre Channel
Freeze and Go 583 Metro Mirror links 122
Freeze and Stop 583 FlashCopy 4, 75, 613
Master Process 582 activating 58
solution highlights 576 and Global Copy 44
Tier 4 580 and Global Mirror for open systems 45
Tier 6 580 and Metro Mirror 43
VolumeSet level 582 apply changes to existing FlashCopy relationship 85
ESS basic concepts 39
Copy Services with DS8000 512 between virtual machines 621
determining volume size 516 between VMware ESX servers 623
ESS 800 commands and parameters used 67
Copy Services 12 commands in the DS front ends 64
create a user ID 513 commit data to target using commitflash 72
interoperability with DS8000 511 comparison of display properties from the DS CLI and
volume address considerations 519 DS SM 84
ESS CLI 716 Consistency Group FlashCopy 48
ESS tasks to migrate 716 data backup system 38
establish a Global Mirror environment with DS SM 355 data mining system 38
establish pair delete existing FlashCopy relationship using the DS
Global Copy 192 SM 89
example display existing FlashCopies using lsflash 70
failover/failback 286 display properties of existing FlashCopy using the DS
examples SM 82
add paths to existing paths between LLSs 164 establish 39
command-line configuration 137 existing Global Copy primary 49
creating Metro Mirror pairs 171 existing Metro Mirror 49
creating paths 160 fast reverse restore 54
delete paths 169 flow 64
DS Storage Manager GUI 230 Freeze FlashCopy Consistency Group 49
establish a Global Mirror environment with DS SM full volume copy 42
355 HP-UX 613
establish Global Copy pairs with DS GUI 234 increment FlashCopy using resyncflash 73
establish paths with the DS Storage Manager GUI Incremental FlashCopy 50
230 initiate background copy for persistent FlashCopy re-
failback 181 lationship 87
failover 179 initiate using Create 80
FlashCopy 102 initiate using mkflash 69
Global Copy and DS CLI 214 integration system 39
Global Mirror and DS CLI 306 interfaces 55, 63
manage Global Mirror environment with DS GUI 369 introduce 256
modify a Global Mirror session 373 limitations with multiple FlashCopies 48
pause a Global Mirror session 372 local 65
resume 177 local commands 68
resume a Global Mirror session 372 local FlashCopies using the DS CLI 66
simplifying Metro Mirror commands 137 managing for ESS 800 532
suspend 175 multiple relationship 48
exportvg 596 nocopy option 42
Extended Long Busy (ELB) 578 operational areas 38
extlongbusy timeout 398 options 47, 55
overview 37
parameters for initial FC using DS SM and DS CLI 81
F parameters used in remote FlashCopy 78
failback 115, 181 performance 92
Metro Mirror 181 Persistent 53
failover 115, 396, 408 planning 58
B to A 276 production backup system 38
Metro Mirror 179 reading from the source 41
failover/failback example 286 reading from the target 41
failoverpprc 331, 343, 409 remote 52, 65
fast reverse restore 54
Index 729
remote FlashCopies using the DS CLI 78 create PPRC paths 215
remote FlashCopy commands 78 determine available fibre links 215
remove local FlashCopy using rmflash 77 establish pair 192
remove relationship 269 establish pairs with DS GUI 234
reset FlashCopy Consistency Group using unfreeze- examples 214
flash 78 examples with DS GUI 230
reset target to contents using revertflash 76 monitoring the copy status with DS GUI 238
resynchronize target using the DS SM 88 overview 188, 202
reverse existing FlashCopy using the DS SM 85 pausepprc 220
reverse restore 54 performance 212
run new background copy for persistent FlashCopy positioning 190
using rmflash 77 primary 49
set an existing FlashCopy to revertible using set- remove pairs 217
flashrevertible 72 remove PPRC paths 218
terminating the FlashCopy relationship 41 resume 220
test system 39 resumepprc 221
using the DS SM front end 80 rmpprc 217
VMware ESX server 617 scalability 212
with other Copy Services 43 setting up environment 215
within a virtual machine 619 state change logic 189
writing to the source 41 suspend 220
writing to the target 41 symmetrical configuration 199
FlashCopy examples Global Mirror 4, 6, 248, 370
create backup 103 add A volumes to session on each LSS 310
create test system or integration system 102 add and remove A volume to existing environment
creating a FlashCopy for backup purposes without 325
volume copy 103 add and remove LSS to existing Global Mirror environ-
multiple setup of a test system with same contents ment 327
102 add and remove Subordinate 329
one time test system 102 add or remove storage servers or LSSs 268
using a target volume to restore its contents back to attributes of synchronous data replication 248
the source 104 basic concepts 252
using an Incremental FlashCopy for backup purposes change tuning parameters 323
104 clear up environment with DS CLI 315
flow close session 318
FlashCopy 64 connectivity to remote site 254
formation Consistency Group drain time 299
Consistency Group 260 Consistency Group formation 260
freeze 120, 579 Consistency Group parameters 261
Freeze and Go consistent asynchronous remote copy solution 251
eRCMF 583 consistent data on B volumes 280
Freeze and Stop coordination time 298
eRCMF 583 create 265
Freeze FlashCopy Consistency Group 49 create FlashCopy with DS GUI 363
freezepprc 393, 398, 429 create Global Copy with DS GUI 359
Full Duplex 396 create PPRC paths from A to B 342
full volume copy 42 create PPRC paths from B to A 339
create session with DS GUI 367
define session 257
G dependent writes 244, 248
GDPS 579 DS CLI 289
GDS 585 DS GUI 291
Global Copy 4–5 DS GUI examples 355
adding and removing PPRC paths 224 eRCMF 293
capacity 212 establish environment with DS GUI 355
changing mode to Metro Mirror 221, 223 failback Global Copy from A to B 344, 352
clear environment 217 failback Global Copy from B to A 339
command comparison 202 failover 276
consistent point-in-time copy 193 failover from B to A 331
converting to Metro Mirror with DS GUI 239 failover Global Copy from A to B 343
create pairs 215
730 IBM System Storage DS8000 Series: Copy Services in Open Environments
failover Global Copy from B to A 349 verify for Consistency Group state 277
failover/failback example 286 verify for valid Consistency Group state 332
for open systems 45 view session’s volumes 371
form Consistency Group 259 volumes 267
interfaces 288 wait for Global Copy first pass to complete 354
introduce FlashCopy 256 Global Mirror Master 383
local site 284 Global Mirror session 253
manage environment 320 pause 372
modify parameters 268
modify session 267, 373
multiple disk storage servers 270 H
non-revertible 278 heartbeat 634
operation 275 help 32
pause 348 DS CLI 31
pause and resume Consistency Group formation 320 Software Information Center 21
pause Global Copy pairs from A to B 349 High Availability (HA) 630
perform disaster recovery testing on D volume 352 HP-UX 613, 615
performance considerations at coordination time 298 hypervisor 631
populate session with volumes 258
primary site failure 275 I
query for Global Copy first pass completion 341 I/O port 357
query out-of-sync tracks for Global Copy 342 IBM TotalStorage Continuous Availability for Windows
quiesce application at remote site 342 585–586
recovery after local site failure using DS CLI 330 introduction 586
Recovery Point Objective (RPO) 296 overview 586, 588
recovery scenario 274 service and ordering 589
reestablish FlashCopy from B to C 337 solution highlights 586
reestablish FlashCopy relationship 350 IBM TotalStorage Rapid Data Recovery 575
reestablish FlashCopy relationship between B and C IBM VDS hardware provider 606
volumes 282 IBM VSS Provider 602
relationship between local and remote volume 255 importvg 596
remote storage server configuration 299 Incremental FlashCopy 50
remove A volumes from session 317 Incremental Resync 437
remove environment 270 Independent Auxiliary Storage Pool (IASP) 635
remove FlashCopy between B and C volumes 319 individual tasks
remove FlashCopy relationship 269 CLI migration 718
remove Global Copy pairs 319 initial disk synchronization
remove PPRC paths 319 Metro Mirror 114
replicating data 244 initial synchronization 130
restart application 282 initiate FlashCopy using Create 80
restart application at remote site 338 Input output processor (IOP) 632
resume 354 integration system 39
return to local site 338 interactive mode
reverse FlashCopy from B to C 334, 350 DS CLI 29–30
revert action 280 interfaces
revertible 278 FlashCopy 55, 63
session 253, 259 Global Mirror 288
simple configuration 254 remote FlashCopy 65
start 345 interoperability 489
start application at the local site 347 DS6000 and DS8000
start for a specified session 311 adding the Storage Complex 498
start session on each LSS 309 creating user IDs and passwords 497
start with Subordinate 314 establish logical paths using DS CLI 504–505
summary of recovery scenario 330 establishing RMC paths 503
synchronous data replication 244 FlashCopy 509
take FlashCopy from B to D 351 hardware requirements 496
terminate 316, 330 licensing requirements 496
terminate and start 324 list available ports using DS CLI 505
terminology 264 manage Global Copy pairs using DS CLI 507
topology 269 manage Global Copy pairs using DS GUI 506
Index 731
manage Global Mirror pairs using DS CLI 508 Master disk subsystem 380
manage Metro Mirror pairs using DS GUI 506 Master Process
manage Metro Mirror pairs using the DS CLI 506 eRCMF 582
microcode levels 496 maximum coordination interval 368, 373
network connectivity 496 maximum time writes inhibited to remote site 373
path creating using the DS GUI 504 metrics 391
storage management 501 Metro Mirror 4–5, 49
updating the DS CLI profile 497 adding capacity in new DS6000s 131
user ID management 500 automation and management 112
volume size considerations for RMC 501 bandwidth 124
DS6000 and DS8000 Copy Services 495 changing options 168
DS6000 and ESS 800 clustering 115
managing ESS 800 FlashCopy 532 Copy Services network components 135
managing Metro Mirror or Global Copy pairs 526 creating pairs 171
DS8000 and ESS 496 creating paths 160
DS8000 and ESS Copy Services 512 data consistency 111
ESS 800 and DS8000 511 distance 125
adding Copy Services Domain 514 DS CLI 136
Copy Services 512 DS GUI 136
volume size considerations for RMC 515 examples 137
introduce FlashCopy 256 failback 115, 181
introduction failover 115, 179
Copy Services 4 failover and failback 115
IBM TotalStorage Continuous Availability for Windows Fibre Channel links 122
586 freeze 120
RMC 5 initial disk synchronization 114
iSeries Access for Windows 643 initial synchronization 130
iSeries Navigator 643 links 121
logical paths 123
LSS design 124
L overview 110
licensing 710 pausepprc 144
authorized level 713 performance 130
charging example 713 resume example 177
links resumepprc 145
Metro Mirror 121 rmpprc 142
local 283 rolling disaster 112
FlashCopy 65 scalability 130
local FlashCopy commands symmetrical configuration 125
parameters 67 traps 705
local site volumes 126
return 284 Metro/Global Mirror 6, 378
logical paths MGM 378
Metro Mirror 123 configuration 382
long busy 398 multiple subsystems 383
lsflash 68, 70, 73, 402 planned recovery 407
lspprc 397 Microsoft Virtual Disk Service see VDS
LSS 268 Microsoft Volume Shadow Copy Services see VSS
add or remove in Global Mirror 268 mkflash 32, 69–70
LSS design mkgmir 290, 346, 390
Metro Mirror 124 mkhostconnect 722
lssession 290, 346, 392 mkpprc 215
LVM 597 mkpprcpath 387
mkremoteflash 229
M mksession 290, 389
man page 32 mkvolgrp 722
manage Global Mirror environment with DS GUI 369 modify a Global Mirror session 267
management console 8 modify a session
managepwfile 26, 498, 514 Global Mirror 373
Master 400 modify Global Mirror session parameters 268
732 IBM System Storage DS8000 Series: Copy Services in Open Environments
multiple relationship Point-in-time Copy see PTC
FlashCopy 48 populate Global Mirror session 258
port IDs
decoding 503, 520
N positioning
nocopy option 42 Global Copy 190
nocp 389 PPRC
non-revertible 278 volume size considerations 515
PPRC links 386
O PPRC path
open systems reason codes 704
AIX and FlashCopy 592 PPRC paths 386
AIX and Remote Mirror and Copy 596 PPRC-XD see Global Copy
AIX specifics 592 primary site failure 275
Copy Services using VERITAS Volume Manager, procedure
SUN Solaris 610 periodic offsite backup 224
Copy Services with Windows volumes 599 production backup system 38
HP-UX and Copy Services 613 profile
HP-UX and FlashCopy 613 DS CLI 25
HP-UX with Remote Mirror and Copy 615 PTC 4
Microsoft Virtual Disk Service (VDS) 604
Microsoft Volume Shadow Copy Services (VSS) 601 Q
SUN Solaris and Copy Services 609 queue full 429
Windows and Remote Mirror and Copy 598 Queue Full condition 578
open systems specifics
Virtual Disk Service overview 604
operational areas 38 R
options reason codes 704
FlashCopy 47 record 389
Out of Sync Tracks 397 recovery
overview 37 Global Mirror scenario 274
Global Copy 188, 202 Recovery Point Objective 378
Recovery Point Objective (RPO)
Global Mirror 296
P recreatevg 595
page fault 632 Redbooks Web site 725
parameters 67, 78, 268 Contact us xxx
Consistency Group 261 reestablish Flashcopy relationship
FlashCopy commands 67 B and C volumes 282
paths remote copy events
add paths to existing paths between LLSs 164 SNMP 705
creating 160 remote FlashCopy 52, 65
delete 169 Remote Mirror and Copy 596, 615
pause a Global Mirror session 372 HP-UX 615
pausegmir 290, 320 Remote Mirror and Copy see RMC
pausepprc 399 remove
Global Copy 220 a Global Mirror environment 270
Metro Mirror 144 remove FlashCopy relationship 269
performance replicating data over a distance 244
FlashCopy overview 92 restart application at remote site 282
Global Copy 212 resume
Metro Mirror 130 Global Mirror session 372
persist 389 Metro Mirror 177
Persistent FlashCopy 53 resumegmir 290, 322
physical connection events resumepprc 221
SNMP 702 Metro Mirror 145
planned outage 637 resyncflash 73–74
planned recovery 407 resynchronize target using DS SM 88
point-in-time copy resyncremoteflash 229
creating Global Copy 193 return codes
Index 733
DS CLI 31 trap 102 703
return to local site 284 trap 200 705
reverse restore 54 trap 202 705
reverse source-target relationship using reverseflash 75 trap 210 705
reverseflash 68, 75–76 trap 211 705
revertflash 76–77, 334, 401 trap 212 706
Revertible 402 trap 213 706
revertible 72, 278, 332–333 trap 214 706
RMC 4–5, 515 trap 215 706
commands 28 trap 216 706
Global Copy 5 trap 217 707
Global Mirror 6 traps 701
Metro Mirror 5 Software Information Center 21
VMware ESX server 626 accessing from DS GUI 21
rmflash 77 accessing using Web browser 21
rmgmir 290, 316, 324 start Global Mirror session 259
rmpprc state change logic
Global Copy 217 Global Copy 189
Metro Mirror 142 Storage Complex 10
rmsession 290, 318 adding 498
Rolling Disaster 577 managing 19
rolling disaster 112 procedure to add 19
RPO 378 Storage Facility Image 8
storage servers
add or remove in Global Mirror 268
S Storage Unit 8
scalability Subordinates 400
Global Copy 212 SUN Solaris
Metro Mirror 130 Copy Services 609
scheduled outage 637 FlashCopy with VERITAS Volume Manager 610
script command mode Remote Mirror and Copy with VERITAS Volume Man-
DS CLI 29 ager 612
script mode suspend
DS CLI 29 Metro Mirror 175
SCSI Queue Full 578 Metro Mirror example 175
session 259, 267 Suspended 398
modify 373 switchover 637
pause 372 swpprc 641, 654
view properties 370 symmetrical configuration
view volumes 371 Global Copy 199
setflashrevertible 72–73 Metro Mirror 125
SFI 8 synchronous data replication 244
showckdvol 721 Synchronous PPRC see Metro Mirror
showextpool 721 sysbas 637, 640
showfbvol 721 System i 630
showgmir 290, 346, 391 Copy Services Toolkit 638
showgmiroos 392 Disk Pool 635
Simple Network Management Protocol see SNMP System i 5
simplifying commands logical partitions 631
Metro Mirror 137 System i5
single-level storage 631 backup node 634
single-shot mode cluster 633
DS CLI 29 cluster node 634
SNMP 701 cluster resource group (CRG) 634
Metro Mirror traps 705 device domain 634
notification external storage 630
overview 702 Hardware Management Console (HMC) 631
physical connection events 702 Input output processor (IOP) 632
remote copy events 705 primary node 634
trap 100 702 recovery domain 634
trap 101 702
734 IBM System Storage DS8000 Series: Copy Services in Open Environments
structure 630–631 Z
subject 632 z/OS Global Mirror 4
zero data loss 378
T
Target Suspended 411
terminology 264
test system 39
tgtinhibit 389
Tier 4 580
Tier 6 580
topology 269
type mmir 388
U
unfreezeflash 78
unfreezepprc 398, 429
unplanned outage 637
user
groups 25
user accounts
DS CLI 25
User ASP 635
user assistance
DS CLI 31
user ID
management 500
V
VDS 604
product components 606
VERITAS Volume Manager 598
view session properties 370
VMware ESX server
Copy Services 616
FlashCopy 617
FlashCopy between 623
FlashCopy between virtual machines 621
FlashCopy within a virtual machine 619
with Remote Mirror and Copy (PPRC) 626
Volume Shadow Copy Services (VSS) 601
volumeaccess 722
volumes
add or remove 267
Metro Mirror 126
VolumeSet level
eRCMF 582
VSS 601
components 601
function 602
W
Windows 598
enlarging extended/spanned volumes 600
extending simple volumes 600
Windows dynamic disks 599
wrkcfgsts 643
WWNN 290
determine for ESS 800 using DS CLI 524
Index 735
736 IBM System Storage DS8000 Series: Copy Services in Open Environments
IBM System Storage DS8000 Series:
Copy Services in Open
Environments
IBM System Storage DS8000 Series:
Copy Services in Open Environments
IBM System Storage DS8000 Series: Copy Services in Open
IBM System Storage DS8000 Series: Copy Services in Open Environments
IBM System Storage DS8000 Series:
Copy Services in Open
Environments
IBM System Storage DS8000 Series:
Copy Services in Open
Environments
Back cover ®