] SHERRYANNE MEYER [ ASUG INSTALLATION MEMBER MEMBER SINCE: 2000 ANUP MAHESHWARI [ ASUG INSTALLATION MEMBER MEMBER SINCE: 2008 AJAY VONKARERY [ ASUG INSTALLATION MEMBER MEMBER SINCE: 1999 How PG&E implemented SAP BI-A and did performance tuning Darin Lemos PG&E Bjarne Berg Director SAP BI Comerit Inc.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
]
SHERRYANNE MEYER[ASUG INSTALLATION MEMBER MEMBER SINCE: 2000
ANUP MAHESHWARI[ASUG INSTALLATION MEMBER MEMBER SINCE: 2008
AJAY VONKARERY[ASUG INSTALLATION MEMBER MEMBER SINCE: 1999
How PG&E implemented SAP BI-A and did performance tuning
Darin LemosPG&E
Bjarne BergDirector SAP BI
Comerit Inc.
Real Experience. Real Advantage.
[
2
Learning Points
See a practical example from PG&E and understand what you can expect on your implementation
Learn how BI-A works
Explore benefits and limitations of the tool
Real Experience. Real Advantage.
[
3
Agenda
Introduction Pacific Gas & Electric Project Overview What is SAP BI-A
Overview Building aggregates Benefits
How we implemented BI-A Sizing the environment Team organization and project plan Testing
We will look what SAP NetWeaver BI Accelerator can—and cannot—do to optimize dashboard performance.
Explore how we PG&E planned, and implemented SAP NetWeaver BI Accelerator.
Examine performance benchmarks for SAP NetWeaver BI Accelerator
Explore the top 10 lessons learned from the project team and see a sample work plan and an organization chart from this project as a reference for your own SAP NetWeaver BI Accelerator initiative…
In this session.
Real Experience. Real Advantage.
[
5
Agenda
Introduction Pacific Gas & Electric Project Overview What is SAP BI-A
Overview Building aggregates Benefits
How we implemented BI-A Sizing the environment Team organization and project plan Testing
Pacific Gas and Electric Company, incorporated in California in 1905, is one of the largest combination natural gas and electric utilities in the United States.
There are approximately 20,000 employees who carry provide natural gas and electric service to 15 million people throughout a 70,000-square-mile service area in northern and central California.
Real Experience. Real Advantage.
[
7
PG&E Company Overview
SAP since 1996
Currently on SAP ERP 4.72, upgrade to ECC 6.0 in 2009
8,500 users for all back office and field operations–Finance/Controlling–HR Org/Benefits/Personnel/Payroll–Employee Self Service: 20,000 employees; 18,000 retirees–Manager Self Service–Supply Chain Management --Materials Management &
Supplier Relationship Management–Equipment/Work Order Management in Plant Maintenance–Business Warehouse
First major go live was a 50,000 person day project in October 2007
Real Experience. Real Advantage.
[
8
PG&E SAP Project Overview
Aug 04 - Nov 04 Feb 05 – Oct 05 Nov 05 – Sep 06
Roadmap Phase
• Vision and Strategy• “To-be” business capabilities• Build business case for change
to enable future business processes and technology
Detailed Design Phase
• Design the foundation for the deployment of new processes and tools
• Design SAP customizations• Design training and performance support
Oct 06 –
Foundational Release: Build, Test, and Deployment Phases
• Build Customizations• Build Procedures and Training• Convert master and transactional data• Assembly Test• Product Test • Deployment
Real Experience. Real Advantage.
[
9
PG&E Project Scope - EAM WRM
Financials(FI)
Account number and budget information
Work Management
(PM)
Asset Management
(PM)
Resource Management & Scheduling Tool
(e.g. CLICK)
Customer web application
Short Cycle Mobile
Supply Chain(MM)
As builts, material reqs, s’ment rules
CUs and design standards
IGISGIS work initiation
Design / Work Order
Initiate Request
Plan & Design
Resource Management and
Scheduling
Perform and Close
WR
M_I
_14
Des
ign
and
asse
t dat
a
Wor
k or
der t
asks
Pla
n an
d fo
reca
st D
ata
Employee skills and availability master data updates
Planned maintenance and inspection
SAP
Sta
tus
chan
ges
& u
pdat
es
Sta
tus
chan
ges
HR
Material availability & Service Contracts
Project Systems(SAP PS)
Time reporting / payroll
Large capital projects
CATS
Contracts / Non Energy Bills
(SD)
Customer contracts
Wor
k st
atus
& c
lose
/ca
ncel
Mark and Locate
Initi
ate
NB
/WR
O
Wor
k
Initi
ate
Cas
e ID
Wor
k ca
ncel
latio
n
WR
M_I
_15
Des
ign
and
asse
t dat
a
Initiate M&L work
FLEETTBD
Lock appt on schedule
User Appt Availability
Graphic Work Design Tool(e.g. ITRON)
Outage
Customer Billing System
GIS
Application for Service
Real Experience. Real Advantage.
[
10
PG&E Project - Scope and Modules
SAP Modules & Versions
PG&E has used SAP for almost 10 years in various forms
R/3 4.7 utilized- (PM, PS, MM, SD,SRM) Modifications to existing FI and HR-CATS functionality Integration with third party applications:
ClickSoftware (ClickPlan, & ClickSchedule), ITRON LD-Pro and ESRI GIS
SAP BI 7.0
Impacted Audience
8,500 (1,500 core EAM, rest for supply chain)
Real Experience. Real Advantage.
[
11
PG&E Project Benefits
Real Experience. Real Advantage.
[
12
PG&E Project BW and History - Performance Needs
Delivering High Performance Analytics to a potential user community of 8,500 users requires special considerations for design, technical architecture and hardware.
I the spring of 2008, PG&E therefore started an examination of BI-A as potential tool for increasing performance of the financial dashboards in SAP BI.
The scope of this effort was primarily finance, but the benefits were also intended to benefit a larger user community.
Delivering High Performance Analytics to a potential user community of 8,500 users requires special considerations for design, technical architecture and hardware.
I the spring of 2008, PG&E therefore started an examination of BI-A as potential tool for increasing performance of the financial dashboards in SAP BI.
The scope of this effort was primarily finance, but the benefits were also intended to benefit a larger user community.
Real Experience. Real Advantage.
[
13
Agenda
Introduction Pacific Gas & Electric Project Overview What is SAP BI-A
Overview Building aggregates Benefits
How we implemented BI-A Sizing the environment Team organization and project plan Testing
Why BI-A? Disk speed is growing slower than other HW components
In-memory data stores
Multi-channel UI, high event volume, cross industry value chains
Application-aware and intelligent data management
Disk based data storage
Simple consumption of apps (Fat client UI, EDI)
General-purpose, application-agnostic database
1990 2009
Architectural Drivers
Improvement20091990
216 Addressable Memory
425x8.51MB/$
0.02MB/$
Memory
166x8.30MIPS/$
0.05MIPS/$
CPU
Technology Drivers
400MBPS
5MBPS
Disk
Data Transfer
80x
100 x10Gbps
100Mbps
Network Speed
264 248x
Source: 1990 numbers SAP AG
2009 numbers Dr. Berg
Physical hard drive speeds only grew by 80 times since 1990. All other hardware components grew faster.
Real Experience. Real Advantage.
[
15
BI Accelerator Works by Indexing InfoCubes in Hardware memory
BI Accelerator has been simple to install and operate for established SAP BI teams
You use process chains to maintain the BI-A engine after each data load
IBM, Sun, HP & Fujitsu have standard solutions ranging from $32K to $300K+
that can be installed and tested in as little as 2-4 weeks (+ SAP license fees)
SAP
BW
Any
tool
32 GB Blades are certified by SAP
Real Experience. Real Advantage.
[
16
BI Analytical Engine’s Query executing priorities
Query ExecutionWithout BI Accelerator
Query ExecutionWithout BI Accelerator
Query ExecutionWith BI Accelerator
Query ExecutionWith BI Accelerator
Information Broadcasting /Precalculation
Information Broadcasting /Precalculation
Query CacheQuery Cache
AggregatesAggregates
InfoProviderInfoProvider
Information Broadcasting /Precalculation
Information Broadcasting /Precalculation
Query CacheQuery Cache
SAP NetWeaver BI AcceleratorSAP NetWeaver BI Accelerator
Aggregates can be replaced with BI-A, while the memory cache is still useful.
Real Experience. Real Advantage.
[
17
BIA Currently reads data from InfoCubes. DSOs & InfoObjects are still read from base/physical tables (even when the InfoObject is indexed as part of master data).
Performance is Near-Linear: Benchmarks
BIA’s strength resides in its near-linear scalability
Performance is measured in terms of:
1. BI-A index creation time
2. Multi-user throughput per hour
3. Average report response time
4. Average number of records touched by each report
Real Experience. Real Advantage.
[
18
Many companies, have spent 2-5 days in BIA training (total for 1-2 administrators) and a work through a single interface:
RSDDBIAMON (Checks available at: RSRV)
Most customers have found BIA Admin to be minimal
The training & admin to run BI Accelerator has proven to
be minimal. Training is included in the SAP courses:
BW310, BW360 and in the online course BW361
In the past, SAP has also offered a 2 day workshop called “BI Upgrade & Accelerator” (WNABIA)
Real Experience. Real Advantage.
[
19
Agenda
Introduction Pacific Gas & Electric Project Overview What is SAP BI-A
Overview Building aggregates Benefits
How we implemented BI-A Sizing the environment Team organization and project plan Testing
SAP has integrated key sizing parameters for BIA in the new QuickSizer. You can give it a try at: http://service.sap.com/quicksizer
Sizing was critical, but good tools were available to help
Customer with existing BI systems and data loaded can get good sizing estimates by running the SAP program available in SAP Note: 917803
Real Experience. Real Advantage.
[
21
The hardware implementation is done by the vendors. Be prepared for lead times on delivery of large systems (Made-to-Order is common).
Some Hardware options we looked at
Operating Systems and StorageSUSE Linux 64-bit, Intel basedMemory based and file storage (no database for Accelerator)
Available Hardware PlatformsHP (HP ProLiant and HP StorageWorks)IBM (IBM Blade Center and IBM Total Storage)FujitsuSiemens (FSC Blade Server and EMC Storage)SUN Microsystems (SUN Blade Server and StorageTek)
Real Experience. Real Advantage.
[
22
We found that the BIA should be sized for critical applications. Most companies use BIA only for Production, while others have a complete landscape
PG&E - Our Initial Hardware sizing (July 2008)Environment Area Recommended size IBM example*Production Blade servers 14 Blades BladeCenter HS21 -8853G6U Production Memory 2x8 GB (2x4) DDR2 total 16 GB 39M5797Production Processors 2 x Quad Core Intel Xeon Processor 2 x Quad Core Intel Xeon Processor Production Processor speed 3.00 GHz+ 3.00 GHzProduction Network cards 2 x Gigabit Cisco cards 32R1760Production External storage Dedicated disks (500 GB+) DS-4800Production File system General Parallel file system (GPFS) GPFSProduction Chassis 14 blades capacity H-series (rack-mount/9U) 88524XU QA Blade servers 14 Blades BladeCenter HS21 -8853G6U QA Memory 2x8 GB (2x4) DDR2 total 16 GB 39M5797QA Processors 2 x Quad Core Intel Xeon Processor 2 x Quad Core Intel Xeon Processor QA Processor speed 3.00 GHz+ 3.00 GHzQA Network cards 2 x Gigabit Cisco cards 32R1760QA External storage Dedicated disks (500 GB+) DS-4800QA File system General Parallel file system (GPFS) GPFSQA Chassis 14 blades capacity H-series (rack-mount/9U) 88524XU Development Blade servers 4 Blades BladeCenter HS21 -8853G6U Development Memory 2x8 GB (2x4) DDR2 total 16 GB 39M5797Development Processors 2 x Quad Core Intel Xeon Processor 2 x Quad Core Intel Xeon Processor Development Processor speed 3.00 GHz+ 3.00 GHzDevelopment Network cards 2 x Gigabit Cisco cards 32R1760Development External storage Dedicated disks (300 GB+) DS-4800Development File system General Parallel file system (GPFS) GPFSDevelopment Chassis 14 blades capacity H-series (rack-mount/9U) 88524XU
Real Experience. Real Advantage.
[
23
PG&E - Our Query Performance Improvments
QueriesCurrent
performance (seconds)
With BIA
Improvement
CS Expense - 5 Year History 169.25 14.54 1064%Charges to Partners via Orders - Units 243.40 21.37 1039%Charges to Partners via Orders $ 171.39 18.28 838%Std Cst Var - PCC Count 136.84 15.72 770%ActPlan vs Target by MWC - Summary 179.09 21.25 743%BU Expense - 5 Year History 201.90 25.23 700%CR - Expense Order Cost In 166.37 24.59 577%Labor Cost In Hours 84.66 14.39 488%24 Month Rolling FC - Amounts 173.60 29.82 482%ActPlan vs Target by MWC - Expense 122.91 21.40 474%SR_RCC_Order Costs by Planning and Actual Orders 171.31 29.86 474%Labor Prod ST Hours 91.01 18.12 402%CWIP - LOB 45.06 9.43 378%Paid OT Hours 83.37 17.55 375%ActPlan vs Target by MWC - Non Earnings 96.28 23.52 309%PCC Labor $/Qty by CE, Plan, Act - Mid-Year 69.32 17.73 291%ActPlan vs Target by MWC - OBS 107.86 28.58 277%CR - PCC Costs Billed Out 163.16 45.07 262%Annual Amt by MWC 100.94 28.18 258%Monthly Run Rate vs Annual EOY 98.82 29.56 234%Monthly Run Rate by Primary Cost Elem 62.01 19.35 220%Activity Price Report (KSBT) 52.59 17.52 200%PCC Non-Labor $ by CE, Plan, Act - Mid-Year 32.99 11.44 188%Month/YTD/Annual Amt by MWC 76.94 28.46 170%Actual Cost Line Items for Cost Centers (Material & Service) 24.48 9.55 156%CR - PCC Costs 62.43 24.49 155%ActPlan vs Target by MWC - Capital 108.49 42.84 153%CR - PCC Costs Billed Out to CC 46.38 18.86 146%Order: Actuals since Inception 25.87 11.09 133%Paid Labor Hours (excludes Overtime) 27.39 11.83 132%SR_PCC Costs_In_Out - Graph 42.83 18.98 126%Actual Cost Line Items for Orders (Other CO Postings) 36.18 16.51 119%Actual Cost Line Items for Cost Centers (Non-PO Invoices) 24.60 11.34 117%Billable Hours 26.55 12.54 112%PCC Std Cst Var Plan - Mid-Year 21.98 10.60 107%PCC Billable Hrs, $ by Actv, Plan, Act - Mid-Year 20.51 10.09 103%
The slowest queries that spent most of their time on the database side benefited the most
Overall 36 of 70 queries were at least twice as fast….
Real Experience. Real Advantage.
[
24
PG&E - Our Query Performance Improvments
58 of 70 queries were at least 50% faster
Before BIA the average query execution took 58.8 seconds, after BIA the average query took 17.9 seconds
(295% faster).
QueriesCurrent
performance (seconds)
With BIA
Improvement
Actual Cost Line Items for Cost Centers (CATS & Order Confm) 21.17 10.74 97%Actual Cost Line Items for Cost Centers (Other FI Postings) 38.13 19.44 96%Summary Removal Costs 11.36 5.90 93%Cost Report by Order/Cost Element (PORR/PORY) 63.62 33.55 90%SR_PCC Costs by Cost In and Cost Out 34.33 18.28 88%SR_PCC Costs by Cost In and Cost Out 34.33 18.34 87%SR_PCC Costs_In_Out - Graph 17.00 9.19 85%PCC Activity Prices 16.66 9.18 81%Spend - Tracked/Not Tracked 14.79 8.23 80%Savings % 8.23 4.58 80%Actual Cost Line Items for Orders (NEBS) 32.81 18.35 79%Actual Cost Line Items for Orders (Material & Service) 30.41 17.12 78%PCC Detail Planning Report (KSBL) 49.62 28.09 77%Actual Cost Line Items for Orders (Other FI Postings) 32.81 18.86 74%PCC Headcount by Month - Classification 21.39 12.34 73%SR_PCC Costs by Cost In and Cost Out - 12 Months 17.66 10.51 68%Actual Cost Line Items for Orders (Non-PO Invoices) 27.19 16.25 67%Actual Cost Line Items for Cost Centers (Other CO Postings) 17.18 10.43 65%Actual Cost Line Items for Cost Centers (NEBS) 24.89 15.33 62%Actual Cost Line Items for Orders (CATS & Order Confm) 24.48 15.59 57%Actual Cost Line Items for Cost Centers (CO Reposting) 18.34 11.90 54%Actual Cost Line Items for Orders (CO Reposting) 25.43 16.55 54%Conforming % 6.68 4.50 48%Activity Type Planning 24.52 16.68 47%Conforming % 6.01 4.23 42%PCC Headcount by Month - Classification (Graph) 15.43 11.58 33%Removal Costs 6.15 4.77 29%Resource Report by PCC/Activity Type (PORV) 39.63 35.42 12%PCC FTE Count by Planning Version - Mid-Year 17.80 15.91 12%CWIP - LOB & Orders 93.04 85.98 8%Savings $ 4.57 4.29 7%Savings $ 4.10 4.02 2%Savings % 3.63 3.94 -8%Conforming % and Savings % 10.37 11.99 -14%
Average performance improvement for workbooks was 363% faster
Current performance
(seconds)
With BIA (seconds)
Improvement
1 ORD205 Actual+Plan by MWC W/Plng Orders 614.63 94.75 649%2 SR_RCC_202 -Order Costs by planning and Actual Orders 171.31 29.86 574%3 ORD225 - Actuals since Inception 25.87 11.09 233%4 Forecast Model (Summary Reports) 438.34 113.01 388%5 ORD418 - Compliance & Savings 58.38 45.78 128%6 ASM001 - CWIP (LOB, MWC, Order) 155.61 106.08 147%7 ORD201 Reports 76.94 28.46 270%8 ORD207 - 5 year historical & FCST 100.94 28.18 358%9 PMG008_24Mo PRJ Fin by Area/Div/RCC/MWC/Ord/PrjMngr/Mo 173.60 29.82 582%
10 PMG005 Reports 63.62 33.55 190%11 ORDPMG001-PORV 39.63 35.42 112%12 ORD230 - Line Items for Order 209.31 119.23 176%13 PCC130 - Line Items for Cost Center 168.79 88.73 190%14 PCC016 - PCC Detail Planning Report (KSBL) 126.73 62.29 203%15 PCC018 - Mid-Year PCC Planning Assumptions for SLSR 179.26 74.95 239%16 PCC116 - Standard Cost & Labor Trend for SLSR Groups 136.84 15.72 870%17 SR_PCC_102 - PCC Costs by In-Out Costs 94.82 47.77 198%18 RCC209A - BU/CS Expense: 5 year history 371.15 39.77 933%19 SR_PCC_401 - Charge-Out Cost Center Type A or E 815.92 165.53 493%20 ORD209 - Plan vs Actual By MWC and primary CE 160.83 48.91 329%
Average run times (same selection criteria) 209.13 60.95 363%
Top 20 Workbooks
Real Experience. Real Advantage.
[
27
Performance Info – PG&E BI-A Indexes build times
232,298,392 records in 23 infocubes were indexed in 90.56 minutes.
On average we were able to build indexes on 2.56 million records per minute
Currently, the BIA performs aggregation and data selection for the query, all other processing is done by the OLAP analytical engine. Not all query processing will be dramatically faster…
In the next SAP NetWeaver release the BIA will handle more of the analytics processing such as “top-5 products sales” which is currently done in the OLAP analytical engine.
SAP BIA is not used when the result set exceeds 3 million records (max. default). When the result set is less, the data is sent as one large data package to the application server (need fast network).
How does it work in reality??
Real Experience. Real Advantage.
[
35
Once you exceed a few hundred critical users and/or 3-4 TB of data you should seriously consider SAP BIA
BIA is becoming mainstream
BIA is no longer something exotic.
Many of the large BI systems have already implemented BIA and
many more projects are under way in Europe and in the Americas.
NikeNike
Some of SAP reference clients
Real Experience. Real Advantage.
[
36
The Future SAP BI-A RoadMap
2009/20102008
SAP NetWeaver BI Accel-
erator
Prior to 2008
Analytic Capabilities (e.g. top n)
DataStore Object Acceleration
Standalone BIA for source-independent acceleration via Business Objects
Agile BI: Local and source-independent innovation space for departments
Analytic Capabilities (e.g. top n)
DataStore Object Acceleration
Standalone BIA for source-independent acceleration via Business Objects
Agile BI: Local and source-independent innovation space for departments
Accelerator for InfoCube Read Times
Low Cost of Operations
High Availability / Backup & Recovery
Accelerator for InfoCube Read Times
Low Cost of Operations
High Availability / Backup & Recovery
Optimized BusinessObjects Webl integration on top of SAP NetWeaver BI/BIA
BusinessObjects Polestar integration: Self-Service BI based on search paradigm
Optimized BusinessObjects Webl integration on top of SAP NetWeaver BI/BIA
BusinessObjects Polestar integration: Self-Service BI based on search paradigm
Real Experience. Real Advantage.
[
37
Agenda
Introduction Pacific Gas & Electric Project Overview What is SAP BI-A
Overview Building aggregates Benefits
How we implemented BI-A Sizing the environment Team organization and project plan Testing
BI-A is a useful tool for performance tuning, but it is not a ‘solve-all’ solution for poorly made queries
It was easy to implement (very quick)
Look at how your queries are executing (i.e. database or application server usage) when estimating benefits of implementing BI-A
Once you exceed a few hundred critical users and/or 3-4 TB of data you should seriously consider SAP BIA
The administration of BI-A is simple and anyone can learn it in a few days
Plan on waiting for hardware, we found that most was made-to-order
Make sure you have redundancies and check that what you order is what got shipped.
Real Experience. Real Advantage.
[
39
Dr. Bjarne Berg's web page -- 85+ presentations, tutorials & articles http://csc-studentweb.lrc.edu/swp/Berg/BB_index_main.htm
Large scale testing of SAP BI Accelerator on a NetWeaver Platformhttps://www.sdn.sap.com/irj/sdn/go/portal/prtroot/docs/library/uuid/b00e7bb5-3add-2a10-3890-e8582df5c70f