Sunteți pe pagina 1din 23

Risk Management @ Electrabel : exploiting SAP BW data for pricing and exposure reporting in Energy Trading

! !

Walter Waterschoot, Risk System Expert, Electrabel

! !

Paul Bruynseels, Cross Application Project Mgr, Electrabel

real world Situation

Compliance

Risk Policy RALCO

Middle Office

Risk Reporting

RMS

Trading Sales

Generation

Modeled implementation

INFLOW: SAS/ACCESS to SAP BW SAS/ACCESS to ORACLE

Pricing functions and methods

OUTFLOW (reporting) ODS (static reporting pdf) SAS/INTRNET(dynamic reporting) Enterprise Guide (ad hoc)

Current market data RMS Current positions Analysis project

Summary table

Instrument values
SAS/Warehouse Administrator (analytical engine) SAS Risk Management

Risk Management System


current market data

current positions

Accounts Payable / Receivable

Mark to Market project

historic market data

Value at Risk project Credit Reports


4

Market data flow


The world outside
mail data sources observations ftp Bloom berg

data currencies; set bloomberg.daily_extract; run; proc sort data=currencies; by ISOCODE; run;

Data collection

current riskfactors

Middle Office
5

Position data flow

Deal Capture System

Deal Capture System

Front Office
data fuel_oil_options; set trading.all_positions; if upcase(type) in (CALL PUT) then run; proc sort data=unit_conversion;

Position collection

current positions

Middle Office
6

IT Challenges

Multitude of data are on different platforms Oracle SAP/BW Excel Integration onto one platform using one software vendor " SAS Access to Oracle " SAS Access to BW " SAS Access to PC files

SAS/ACCESS to SAP BW

Where used?
Risk Mgmt
To access financial data coming from SAP/R3s FI-CO environment daily ! uploaded into BW ODSs To access Trading Master Data coming from Oracle database made available in BW

SAS/ACCESS to SAP BW
!

Features

Uses SAS/ACCESS to R/3 technology underneath Accesses Infocubes, ODSs and PSA tables Works on non-Windows platforms High performance -> good for mass extraction Wizard to load BW metadata into SAS Wizard to read BW InfoCubes, ODSs and PSA tables into SAS Wizard to export OLAP metadata to SAS/EIS
9

Structure of InfoCube in SAP BW


! !

BW stores InfoCubes in Snowflake Schema:


1Fact table n Dimension tables n SID tables n Text tables n Master tables
T S M T M T S D T S D S F D T S D S M
10

T M

Extraction of InfoCube
!

Reduces snowflake schema to star schema:


I I

1 fact table n dimension tables SAS Data Sets Star Schema Tables D T S M D T S M D D F D

SAS Views to BW Snowflake Schema Tables T M S D M S T D F D S S T MT M

SAS Formats

11

SAS/ACCESS to SAP BW
Architecture Details Metadata

SAS/ACCESS to SAP BW SAS/MDDB BASE Extractor


Meta data Repository

SAS/ACCESS to R/3 SAP BW


Meta data Repository

12

SAS/ACCESS to SAP BW
Architecture Details InfoCube

SAS/ACCESS to SAP BW SAS/MDDB BASE SAS/ACCESS to R/3 SAP BW


InfoCube Star Schema

Extractor
Meta data Repository

13

SAS/ACCESS to SAP BW
!

PROS

Easy interface Accesses Infocubes and ODSs High performance with ODSs especially when dedicated server used # good for mass extraction
I

CONS
No field selection possible No where clause possible When extracting complex infocubes performance tends to slow down ! processing was introduced and tested but practically no gain (SAP BWs snowflake complexity)
14

SAS/ACCESS to R3 to access SAP BW


!

PROS

Easy interface Accesses PSA tables easily High performance especially when dedicated server used # good for mass extraction Field selection possible Where clause possible No internal BW processing since only staging area used (very fast)

15

SAS/ACCESS to R3 to access SAP BW

CONS

16

SAS/ACCESS to R3 to access SAP BW

Where used?

Marketing and Sales

To access invoice data coming from several non-SAP environments uploaded into BW PSA tables

17

Overall Remarks

Ordinary userids as defined in BW for end users cannot be used to access BW through SAS Access
The SASCPIC userid that has to be defined to be able to access BW with SAS Access cannot restrict data access to the infoarea or data target level (its all or nothing)

18

Tips and Tricks


If data stored in BW as data repository only without any cubes built on top for query functionalities then use PSA instead
Rather extract data from ODSs instead of Infocubes (less complex # faster) Define several SASCPIC userids in BW and dedicate each of them to different processes, depts, groups to avoid interference To restrict data access, download restricted metadata datasets to different metadata files with restricted access Keep it simple

19

Some download times


ODS (with dedicated server)
# obs. 2.983.440 2.460.813 217.008 size 1.909 Mb 2.524 Mb 267 Mb # vars 88 15 199 Downlaod Time 32 13 3 41 5 14 Rate 1 Mb/s 11 Mb/s 1 Mb/s

PSA (from client workstation)


# obs. 1.697.563 4.415.689 size 92 Mb 2.594 Mb # vars 3 64 Download Time 3 33 47 12 Rate 0.5 Mb/s 1 Mb/s
20

Benefits
Builds on previously made investment Harnesses the power of SAS Complements BW functionality Metadata driven extraction without programming Offloads analytical processing to specially tuned analytical SAS data mart Only product on the market to surface InfoCube data Not limited to Windows platforms -> scalable Prerequisite for data mining
21

Certification
!

SAS has an SAP-certified software interface for Business Information Warehouses OLE DB for OLAP (ODBO) interface

22

S-ar putea să vă placă și