Sunteți pe pagina 1din 3

Mapping Services, Systems and Servers using the

CMDB
David Cuthbertson, Square Mile Systems

Understanding the dependencies between IT components is critical to managing change


effectively. The relationships between services, software and hardware should also be
known when creating risk and recovery plans. However, the creation and maintenance of
service maps is often problematical due to scope, the amount of manual effort involved in
data capture, the limited skill sets available and a suitable data repository, or CMDB.

Why Create and Maintain Service Maps


Many organisations get by without mapping their service infrastructure, but there a growing need to create a
CMDB (or equivalent) if you want to get processes more integrated. We know that trying to document
complex IT systems is painful and involves lots of manual effort in data capture and upkeep, but you will
have to create service maps at some stage.
CI Hierarchy
The hierarchy diagram on the right shows
typical types of layers used for service
Business Process CRM, Accounts Payment, HR
mapping. We need to understand the
dependencies between and within layers Service Payments, Customer Complaints, Payroll
when managing change to critical services.
In addition we may map documents and Systems Collections of Software Applications
other data to the different Configuration
Items (CIs) with greater levels of detail. Software Specific applications, data or databases

For many, this model is too complex for Software Infrastructure DBMS such as Oracle, SAP
general use so often the layers are bundled
together. The systems layer may comprise Virtual Infrastructure Virtualised Servers, PCs, Storage, VLANs
all software, databases, and external feeds
Hardware Servers, PCs, Networks, Storage, Networks
for instance. You have to make a design
choice – too many layers in the CMDB Environment Locations, Cabling, Power
results in more complexity to manage, too
few and it is of limited value.

There are many practical reasons why you create service maps.

1. It’s the easiest way to record relationships so that they can be input into a CMDB.
2. The relationships between CIs in an existing CMDB can’t be validated easily without a service map.
3. Without services and component dependencies mapped, IT teams will often categorise changes and
impacts inaccurately, resulting in unnecessary disruption and inconsistent reporting. You can’t see
“hotspots” on a service map if the map does not exist.
4. Reacting to an incident or unforeseen event is made more difficult if you can’t determine
dependencies and potential risks of any actions you intend to take (or worse, have already taken).
5. Service reporting becomes of limited use as incidents and problems can’t be assigned easily as far
down the hierarchy tree as possible to reflect the root cause.
6. Problem management is time consuming and less effective if you have to create a service map for
each problem review! This is one of the reasons why there are white boards in incident rooms.
7. Service continuity, availability, capacity, financial and other management processes become too
difficult and costly to manage easily as single points of failure; duplication and other factors are too
difficult to comprehend. So each team creates its own knowledge base not correlated with others.
8. Information on major incidents or problems can be “distorted” by individuals as detailed knowledge of
system interaction is released economically to support opinions, commercial or political objectives.
9. There is a contractual obligation, or specific objective to show evidence of how you map services
and maintain them. Somebody wants to see the evidence, so it has to be done!

In addition to these, you may have your own requirements, ranked with priorities.

© 2007 Square Mile Systems Limited Page 1 of 3


What types of Service Maps are there?
There are many ways to look at service mapping.
“Top Down” Dependency Mapping
Different types of maps are often found within service Security & Risk Planning
desks and other support tools, so a decision must be Consolidation Risk
made on the type of view you need to deliver first. Are Systems Architecture
you looking for a “top down” service map – ie starting Availability Planning
at a service and looking at underlying or supporting Business Continuity
Single Points of Failure
CIs.

Or do you need a “bottom up” view starting at a The


component such as a server and looking at the CMDB
software and services affected? The picture on the
right emphasises typical uses of the CMDB and the
type of mapping best suited.

When you collect the information it may in different Impact Analysis


Incident Management
formats, with the benefit of the CMDB that they will all Crisis Management
be normalised into a common data structure. When Change Control
you want to understand how CIs are linked then the “Bottom Up” Project Planning
presentation may also take various forms. Each has Incident Recovery
its own benefits, with some not scaling well for the
bigger, or more complex environments. CMDB Views

Below are some examples of different methods used to display relationships between CIs, all providing
different types of understanding from a common set of CI data.

User 1 H/W 1
Payroll
Payroll Service
Software
User 2 H/W 2
Software Pay one
Pay one Overtime
Overtime S/W 1 Payroll DB 1
BACs pay
BACs pay
Hardware S/W 2 SLA 1

Hardware SVR001
SVR001 SVRAccts S/W 3 SLA 2
SVRAccts

Tree Hub & Spoke


List
Typical Service Map Presentations
ISA Pay ment Func tion

Payroll ISA Pay ment Reques t


Digita l Sour ce s Ing est
Catal ogin g
ISA Pay ment Reques t Handling Client Funds Trans fer
Ph ys ic al Arc hi ve

Cus tomer Serv ic es Z-M i cr os yste ms Av id Vir ag e Vi d eo lo g g er Fi na l C ut Pro So ny EVTR


JAVA 01 Arte si a fo rD AM 6 .0
Vi ta Re posi tor y

O rac l e M as ter
SAM -FS
Citrix Client ISA Pay ment Work flow Client Trans ac t BACS-IP BACS-Sc hed Claims Proc es s ing

Sun Fir e V89 0 So ny Pe ta s it e QFS V4.3 Store dg e 69 20


Ph ysica l Sour ces Playo ut

S/W 1 S/W 2 S/W 3 Tr ans code


Sun Fir e V24 0Z

Work flow Pay log Audittrac k BACPAY Funds Mov e Claims Proc es s ing Qu an te l L ei tch O mn eo n Pi nn ac le Se ac ha ng e Tho ms on G ras s val l ey

Fli p Fac to ry

D i gi ta l Arc hi ve

Sun Fir e V40 0Z

Orac le FWS_03 SQL FWS_04 Orac le LP1 Orac le LP2 Loc al Bro adc ast & Cab le Digita l Wa ter mar k
Au tom ation

DB 1
Ne xtamp

Citrix Serv er Rad io Fi le Serve r Ai r Cl ie nt De vi ce se rve r M ed ia Cl i en t

UK_VWBIRM001 UK_VWBIRM002 UK_VWBIRM004 LPAR1 LPAR2


Su nfi re V40 Z

Ha rri s

Ed it
Tap e
Te lev isio n Netw or ks Tr affic
Serv ic e Des k UK_BIRM_BLADE_01 UK_BIRM_BLADE-02 Cheque Printer HO_BACRT01 BACS IP-VPN Z800 2065
UK_BHAM-010301
Web Ser v ices

MF_RT01 Chu n ker


M yers Protra ck 5 .0
MFLAN JAVA Web Se rvi c es

H/W 1 H/W 2 Front Des k HO Sec ure LAN FW_FWS_01 MF_RT02


I nter ne t Acces s

Sun Fir e V49 0Z


J AVA0 2

SW_FWS_05 BT Global MF_SW01 MF_SW02


SW_FWS_01

Heirarchy Complex Hierarchy Data Flow

The top three views are relatively easy to automate with software as the display structures are simple, but
are not suited to large numbers of CIs, or multiple relationships. The bottom three provide more information
about service and system dependencies, but are more manual in nature. Many of the more modern CMDB
systems try and automate the production of these service maps, though there is often a manual tidy up
required to enable comprehension. After all, a map is designed to suit a need and we may need just an
overview, a detailed technical view or a mixture of both. And we still have to get the base data to begin with!

© 2007 Square Mile Systems Limited Page 2 of 3


In the authors experience, the complex hierarchy view should be the primary aim as the information is suited
to multiple uses. For instance, the same service map will help teams categorise changes at the best level. It
will enable a service continuity manager to determine which CIs are delivering the critical parts of a service.
Billing for services can be easily determined depending on which CIs are shared or dedicated to business
groups. Not an issue if you have only a few servers, but a definite benefit if you have hundreds of servers
spread across multiple equipment rooms. We can more easily determine capacity, recovery and the impact
of failures with a complex hierarchy map, than with the list and tree views.

The Service Mapping Project


The ideal way to start is to have available details of your services, hardware, software and mapping
standards so that the project team can focus on understanding relationships between CIs. For many, these
details do not exist so a more pragmatic approach is required – one that delivers some tangible results in a
fixed timescale.

Our recommendation is that you start with a few key services to help drive the definition of standards,
naming and display. It also shows up the quality and depth of operational data, making it easy to focus
attention on what should be audited and how. Auto-discovery packages can help initially with device details,
but as much of the service mapping data is manual in nature, they are of limited use.

As the majority of services are mapped, the ownership of each service map should be assigned, often to a
business service owner or service performance manager. It becomes their responsibility to regularly validate
the service maps for which they are responsible. The configuration manager, or a team colleague should
have the skills and tools to generate service maps from the underlying CMDB, supporting validation needs
as well as specific project or management requirements.

The end goal is to have a populated, maintained CMDB from which service maps can be derived and
updated to help the various process teams in their roles. As service desks are not primarily selected for their
ability to present service maps, it often requires some reporting software or middleware to extract CMDB
data, so service maps of the various types can be easily produced by visualisation tools.

A newer technique that is being increasingly adopted is to link service maps to incidents, problems and
changes to reduce the discovery time during incidents or change reviews. It then becomes easy to see
where there are “hotspots” or potentially conflicting changes across a large environment. It does of course
assume that you already have a service map to link to…

The Author
David Cuthbertson is a founding director of Square Mile Systems, a UK computer services company focused
on configuration management solutions. He is a regular industry speaker at ITSMF events on best practices
and the application of configuration management techniques to ICT and data centre infrastructure. He is the
current chairman of the BCS Service Management Specialist Group (SMSG).

Worldwide Headquarters (Europe & US)


Square Mile Systems
3 Church Street, Cirencester
Gloucestershire, United Kingdom GL7 1LE
Tel +44 (0)870 950 4651 Fax +44 (0)870 751 9268
www.squaremilesystems.com

Asia Pacific (including Australia)


Tel +61 (0)417 231726

© 2007 Square Mile Systems Limited Page 3 of 3