teched2009_ALM107

1

ALM107

Running a Sizing Project from Blueprint to Upgrade

Performance & Scalability, SAP AG

Disclaimer

This presentation outlines our general product direction and should not be relied on in making a purchase decision. This presentation is not subject to your license agreement or any other agreement with SAP. SAP has no obligation to pursue any course of business outlined in this presentation or to develop or release anycourse of business outlined in this presentation or to develop or release any functionality mentioned in this presentation. This presentation and SAP's strategy and possible future developments are subject to change and may be changed by SAP at any time for any reason without notice. This document is provided without a warranty of any kind, either express or implied, including but not limited to, the implied warranties of merchantability, fitness for a particular purpose, or non-infringement. SAP assumes no responsibility for errors or omissions in this document, except if such damages were caused by SAP intentionally or grossly negligent.

© SAP AG 2009. All rights reserved. / Page 2

2

Agenda

1 Sizing and the Product Lifecycle1. Sizing and the Product Lifecycle

2. Sizing Basics

3. Sizing Tools and Results

4. Follow-On Activities and Related Topics

5. Sizing Tips and How-TosDivider headline


Sizing Questions May Come in Different Flavors

Can SAP software payroll 3+ million l d i i l th“

Typical sizing questions Definition of sizing

The iterative translation of employees and pensioners in less than

two hours?

When I check your sizing Portal, there is no sizing guideline for

FIN-FSCM-TRN. How can I do a sizing?“We are running on 4.6C and want to “

business requirements into hardware requirements

Feasibility is only very loosely tied to sizing


upgrade to ERP2005.

We have 30,000 business partners and 60,000 sales orders with 50 line items per month. How much hardware does

our SAP application require?“

3

Different Times, Different Phases, Different Goals of Sizing

Sizing is an iteration, it takes place in different phases of a project

Project Preparation

Business Blueprint Realization Final

PreparationGoing Live & Support Upgrade Migration Business

UnitsFunctionalChanges

Very early to plan hardware expendituresA few months before live start to verify assumptions

Re-define the overall performance requirements

A few weeks before go live to check hardware capacity

During production stages to ensure operations and verify/adjust estimations made earlier.

”Trigger events” include:– Upgrade database, operating system,

SAP application– Reconfigure system landscape


p y g y p– Change business process– Successive rollouts: more users or

other load

Possible Definitions for Different Types of Sizing

Hardware Budget Sizing Advanced Sizing Expert Sizing

Smaller companiesVery simple algorithmsAssumptions likelihoods

Medium to large companiesThroughput estimatesQuestionnaires formulas

Large or complex projectsAdditional guidelinesCustom calculations

GoLive

Assumptions, likelihoodsLevel setting of projectRisk identification

Questionnaires, formulas Usage of standard toolsFocus on core business processes

Custom calculationsAnalysis of custom coding Custom sizing guidelines

All projects All projects All projects

Initial Sizings

Re-Sizing Delta Sizing Upgrade Sizing


All projects

SAP system monitorsGoal: Extend an existing system by load

E.g. by volume100 additional users who'll do the same as the current productive ones

All projects

SAP system monitorsGoal: Extend an existing system by functions

By different functions, e.g. you are live with CRM and want to add SCM

All projects

SAP system monitorsSAP NotesGoal: Upgrade SAP software

Production Sizings – whenever there is a change in throughput, sizing must be done

4

Some Factors That Influence Sizing

System Settings

ParameterizationInterfacesSecurity settings

Set up of business processesOrganizational

CustomizingHW Platform

Processor technologyDisk technology

SAP Software

ReleaseOLTP or OLAPIndustry solutions

UnicodeA2A, B2B scenario

structuresBusiness process design

Network technologySystem infrastructure

User Behavior

ConcurrencyAvoiding data

Data Volume

Time for volume

Disk GrowthCustom Coding, 3rd Party

Performance

Customer profile


yLAN/WANInternet/intranetActivity, e.g.

*-SearchEfficient navigation

gArchiving strategiesInformation Lifecycle Mgmt.

processingBackgroand processing, parallel jobsReportingData distribution

impactScalabilityBusiness process design

Technology Partner CustomerSAPResponsibility of

Customer profile

Agenda

1 Sizing and the Product Lifecycle1. Sizing and the Product Lifecycle

2. Sizing Basics



5. Sizing Tips and How-TosDivider headline


5

Universally Applicable Statements on Sizing

20% of the transactions and scenarios create 80% of the load”Reliable sizings can only be done withReliable sizings can only be done withScalable application softwareScalable hardware and technology componentsMeaningful and nearly complete volume throughput or active user

information

”

Sizing results in overall hardware requirements in KPIs Processing power (CPUs, cores, processors, threads)”


Disk sizeDisk I/O Main memoryFront-end bandwidth requirements

Landscaping and configuration are related, but separate tasks”

Details on the Sizing Process Steps –Prerequisites for the Sizing Process

ThereforeSAPconducts performance and scalability tests on all applications in single user mode for each

20% of the transactions and scenarios create 80% of the load”

release

SAPcreates test kits to allow volume testing of major applications

Hardware vendors and technology partnersrun SAP Standard Application Benchmarks to prove the scalability of their hardware

SAP

Reliable sizings can only be done with

Scalable application software

Scalable hardware and

”


SAPpublishes these benchmark certifications at www.sap.com/benchmark on behalf of the partners

technology components

6

Small test system

Dimensions of internal and external scalability Effects of scalability

Dimensions and Effects of Scalability

Small data source

One user

Thousands of usersTerabytes of data

Size of data

Number of processed

objects

Concurrentusers

Number of servers / cores / threads

Scalability


yHigh throughputLarge objectsGlobally distributed usersHardware adaptivity

Many servers

Number of objects in the DB

Number of parallel jobs

Linearity of CPU Measurements

Linear scalability Stable CPU time while response time increases under load

50

60

50

60

No scalability Load-optimized scalability

0

10

20

30

40

0 100 200 300 400 500 600

90000

0

10

20

30

40

1 5 10 20 50 100 200 500


0

2

4

6

8

10

12

14

16

0 100 200 300 400 500 600 line items

0

10000

20000

30000

40000

50000

60000

70000

80000

90000

0 10 20 30 40 50

CPU

tim

e in

ms

Application Database

7

Ensuring Scalability With Performance Tests –Approaches

Single User TestSmall test system (QA, development), one user

Quality and implications of accesses to persistence layerLinear resource consumptionParallel processing mechanisms, load balancingMemory usage and memory leaksDisk requirementsFront-end network load

Analyze & measurescalable behavior

Performance predictions/sizing


Volume TestEquivalent to multi-user test, stress test, load test, benchmark

for high volume environment

Verify assumptions and ...

Assemble-To-Order (ATO)Cross Application Time Sheet (CATS)Financial Accounting (FI)

SAP ERP

E-SellingInteraction Center (IC)

SAP Customer Relationship Management

Benchmarks Covering SAP Business Suite

Industry standardCPU-drivenUsed since 1991Delivers SAPSS b dFinancial Accounting (FI)

Human Resources - Payroll (HR)Materials Management (MM)Production Planning (PP)Sales and Distribution (SD)

Business IntelligenceBusiness Information Warehouse (BW)

SAP NetWeaver

Warehouse Management (WM) Advanced Planner and Optimizer (APO)Demand Planning (DP)Prod. Planning/Detailed Scheduling (PP/DS)Supply Network Planning (SNP)

SAP Supply Chain Management

P j t S t (PS)

SAP Product Lifecycle Management

Sub-second response time as of 2009 In Memory

Computing


Data Mart (BI-D)Mixed Load (BI-MXL)

Enterprise PortalEmployee Self-Service (EP-ESS)People-Centric CRM (EP-PCC)

SAP Business Solutions & Platform

SAP for Banking – Transaction Banking (TRBK)SAP for RetailSAP for Utilities - Customer Care & Service (ISU/CCS)

Industry Portfolios

Project System (PS) I/OCPUMemoryCPU

Memory

8

User Interaction Steps SD Standard Application Benchmark

0. Logon 10. Choose Enter1. Main screen 11. Call /nVL02n (Change delivery)2. Call /nVA01 (Create customer order) 12. Choose [F20] (Posts goods issue)

Steps 2 to 16 Repeated n times 15 steps take at l t 150 d3. 1st screen 13. Call /nVA05 (List orders)

4. 2nd screen (with 5 items) 14. Choose Enter5. Choose Save 15. Call /nVF01 (Create invoice)6. Call /nVL01N (Create a delivery) 16. Choose Save7. 1st screen 17. Call /nend8. Choose Save 18. Confirm log off9. Call /nVA03 (Display customer order)

least 150 seconds

Business aspectOne run (steps 2 to 16) corresponds to the selling of 5 line items


Multi-User Simulation

High Load PhaseRamp

up

Ramp

down

012345678

Time

# U

sers

Definition: SAPS

S APA pplication

Performance baseline defined via theoretical reference machine* Can be adapted to new technologiesEnables the comparison of different platforms, client/server configurations

*[Haas & Zorn 1995] Haas, M.; Zorn, W.: Methodische Leistungsanalyse von Rechensystemen.Reihe: Handbuch der Informatik 2.6,Oldenbourg, München, Wien, 1995.

A pplicationP erformanceS tandard

p p , g(2-tier, 3-tier, multi-tier)Applicable to SAP Standard Application Benchmarks: business processes anderlying the SAP application software do not change significantly Hardware-independent unit of measurement SAPS is derived from Sales & Distribution (SD) Standard Application Benchmark


** 6,000 dialog steps and 2,000 postings or 2,400 SAP transactions=̂

2,000 fully processed

order line items/hour**

100 SAPS

SD Benchmark

9

Example – Certification 2009005

© SAP AG 2009. All rights reserved. / Page 17Source: www.sap.com/benchmark

SAP SD Standard Application Benchmark Results

Sorting options

Display certificate

ExampleRequired from sizing 2,000 SAPS


10

Details on the Sizing Process Steps –Executing the Sizing

ThereforeSAPprovides tools such as Quick Sizer to enable customers to size using the appropriate

Reliable sizings can only be done withMeaningful and nearly

l t l

”customers to size using the appropriate information on the planned system usage

Customersfill in these tools with as much reliable data as possible in an iterative process.

complete volume throughput or active user information

20/80 Rule20% of the transactions and

scenarios create 80% of th l d

”


the load

Details on the Sizing Process Steps - Using the Sizing Results

ThereforeSAP’sstandard sizing tools provide overall sizing results in an hardware-independent format

The result of sizing: Overall hardware requirements in KPIs ”

results in an hardware independent formatProcessing power is highly platform-dependent, and SAP cooperates with many different partnersFocus on SAP standard

Hardware vendorstranslate sizing results into their own hardware termsCustomers

Processing power Disk sizeDisk I/O Main memoryFront-end bandwidth

requirements


Customerscan cross-check hardware requirements using results from SAP Standard Application BenchmarksHardware vendorssuggest configurations and hardware landscape

Including possibilities for virtualization

Landscaping & configurationAre related, but separate

tasks”

11

Three-Party Collaboration Model

Three parties collaborate in the benchmarking and sizing process

ContributionsCertified benchmarks scalable

h d

Expectations from benchmarking and sizingOptimal performance

Hardware Vendors Customer

hardwareDifferent configurations together

with technology partnersPerformance studiesCustom load tests in collaboration

with customersService level agreements

Optimal performanceSuggestion for hardware

configuration

Contributions Response time requirementsThroughput requirementsB#usiness data input

SAP

Sizing RecommendationCPU (SAPS) Memory (GB)Database space (GB)Disk I/O operations per secFrontend bandwidth mbps


ContributionsDevelopment and provision of

benchmark toolkitsRegression testing for new releasesStandard sizing guidelines as part

of quality assurance processSizing verification processesQuick Sizer and other tools

Risks in a Sizing Project

INCOMPLETE INPUT DATAThe challenge in any sizing project is to obtain sufficient usage informationOften caused by communication issues

1Often caused by communication issues

Project team from system administrationImplementation team knows the core processes best

When there is not sufficient data available, assumptions must be made – and documented

ASSUMPTIONS ARE NOT VERIFIED

While it is perfectly ok to work on assumptions you must ensure that a verification process is included in the project plan

2


verification process is included in the project plan

CUSTOM CODING & SPECIAL DATA CONSTELLATIONS

Are very hard to predict: Make sure there is a verification processMake sizing measurements, if required

3

12

Agenda

1. Sizing and the Product Lifecycle

2. Sizing Basics



5. Sizing Tips and How-Tos


Some Key Performance Indicators

CPUProcessing times of business transactions or tasksCost factor: Number and processing power of servers

Disk sizeDisk I/O

Data that resides on the databaseFile read and write activity to storageCost factors

Backup/recovery depends on size of databaseStorage capacity

Allocated to a user or background process


Memory Garbage collection, acceleration, planning capabilities, buffers, cachesCost factor: Physical memory slots

Front-endNetwork

Load

Transferred amount of dataNetwork time and roundtripsCost factor: Leasing bandwidth

13

Sizing Information Platform: http://service.sap.com/sizing

GuidelinesGeneral proceduresApplication-specificApplication specific

ToolsQuick SizerOthers

Training opportunities


Different Standard Sizing Methods and Tools

Sizing in overview tablesSimple algorithms with many assumptions

Detailed Sizing Questionnaire Without Formulas

For structured questions

Online sizing tool Quick SizerSizing in formulasSi l l


Simple or more complex

14

Quick Sizer, SAP’s Online Sizing Tool

CharacteristicsStructured sizing questionnairesInput for

Initial sizingInitial sizingGoingLive Check

Hardware vendor contact listOn the Service Marketplace

Scope Facts & Figures


Key applications fromBusiness SuiteIndustriesSAP NetWeaver

Sizing by users and/or by throughput

Available online since 1996Free of charge300,000+ projects Avg. 45,000 new projects per year30,000+ different users

Process for a Quick Sizing – Overview

PreconditionsAccess to the Service MarketplacePlanned user and volume information

For an initial sizing recommendation, proceed as follows

1 Call http://service.sap.com/quicksizing

2 Create a sizing project with the relevant information, such as number of users

3 Get an initial sizing result for CPU, disk and memory

4 Possibly apply additional guidelines on top


5 Check for sample configurations at www.sap.com/benchmark

6 Provide hardware vendor with Quick Sizer project name (and additional guidelines, if desired)

15

Design Guidelines: Two Methodologies of the Quick Sizer

Quick Sizer Sizing based on users Sizing based on throughput

UsersLow, medium, high activity Application componentsAssumptions for load on CPU and disk

Result Average CPU (SAPS)Disk sizeDisk I/O

Business throughput Actual objects/transactions/scenarios

Considers time periodsResidence timeAverage working dayPeak load time frame

ResultAverage and peak CPU (SAPS)

Sizing based on users Sizing based on throughput


Memory

ScopeSufficient for small projects with low user numbersSome application do not require user sizing (e.g. Exchange Infrastructure)

Average and peak CPU (SAPS)Disk size and growthDisk I/OMemory

ScopeFor larger or complex projects

DEMO

16

Structure of the Quick Sizer

Project-wide informationWork daysStatus

2

StatusOwnerMethodLinks for hardware vendorsDocumentationDisclaimer


QuestionnaireHardware infrastructure (optional)Per solution / key capability1

3

NavigationSolutionsKey capabilities

Navigation – Limit User Access

If you check the flag “Limit user access“, only you as project owner can grant other persons access to the project data:

Enter the user-IDs of these persons and choose the

1

p“Add user” button (then you see these user-IDs in the drop-down listbox)To delete persons from your access list, set the cursor on the user-ID in the drop-down listbox and choose „Delete user“

In your project overview list protected projects are marked


If you uncheck the flag, your project can be accessed by everybody who knows the customer number and the project name

17

The search is version dependent. The program searches for your search string across the visible and invisible parts of the Quick Sizer. The results are displayed in the navigation tree.

All result lines are marked with an information icon. If the string is found in the visible parts the search

Navigation - Search

1

If the string is found in the visible parts, the search string result will be presented bold and enlarged.If the search string is found in the invisible parts, the entire result line will be presented bold, enlarged, and in italics.If a result only partially contains the search string, only the search string is marked bold and enlarged.

If you check the flag “Match whole word”, words which only consist of the searched word are displayed. For example, if you enter “HR” without the flag, you also get “… synchronized …” in your search results. If you check the flag “Match whole word”, these kind of results will not be displayed.

“ “O “ f


You can also choose between “Phrase”, “Or” or “And”. If you choose“Phrase”, you only get the results which contain the exact search string. For example, if your search string is “Sales & Service”, you get results such as “Sales & Service”. “Or”, you get all results which contain one or more words of your search string. For example, if you search for “Sales Service” you get all results which contain either sales or service or both. “”And”, you only get the results which contain all your search strings. For example, if you search for “Sales Service”, you only get the results which contain both words.

Navigation

To open a questionnaire, click on the text

If you click on the arrow, the list is displayed:Links to sizing guidelines that are not included in the

1

Quick Sizer

Meaning of the icons:Green: questionnaire filled in, no errorsYellow: no entries in questionnaireRed: questionnaire filled in and contains errors

If errors, no results are calculatedIf you open a project with errors on specific

i i h i i h h k


questionnaires, the navigation show the key capabilities with errors

Light bulb: Current questionnaire

18

Project Header Information

2

Functions in the project headerSaving of sizing data

There are two methods for saving your input data:– Manually, by using the save button– Automatically, by navigating from one questionnaire to the next

Create print page for each pageCalculate resultSet project to ‘GoingLive’ and ‘Final’


Set project to GoingLive and Final Only active, if entries are made in this project

Documentation on using the Quick Sizer in generalDirect access to hardware vendors sizing informationDisclaimerChange the number of work days per year

Influences CPU, memory, and I/ODefault: 220 days

Details on Project Header Information: Documenting Sizing Projects: Print Files

Print files


t esFrom every page in the Quick Sizer, you can create a print file

Print directlySave to disk

Helpful for status documentation

19

Details on Help and Online Documentation

DocumentationGeneral documentation on Quick Sizer

HelpPer questionnaire

Additional sizing guidelines


Per questionnaireCollection of useful links

Structure of Sizing Questionnaires

3

3a Questionnaire-wide information

3b User sizing

3c Throughput sizing


20

Details an Questionnaire-Wide Information

3a

Changing average workday and peak timesDefault workday: 9-18, default peak: 12-13You can modify these entries at questionnaire level

The average workday entry will change for all average sizingsThe default peak will be changed for lines without entries and new lines

Ch k i t

Long headersYou can choose between long and short headers

Delete QuestionnaireYou can delete all information you entered on a questionnaire by using this button

LinksHelp: help on sizing elements of this questionnaire


Check inputYou can check the questionnaire for entry errors

Use Default ValuesYou can reset to default values if you changed them before (e.g. for Portal)

Add. sizing guidelines: collection of related topics

Side Step: Error Handling 1/3

Input errors such asMistyping - e.g. give in letters instead of numbers – will be automatically cleared

There are basically two types of error handling:There are basically two types of error handling:Non-requested

Upon leaving the questionnaire (indirect save)When you save the questionnaire data

Requested Use check function


21


You can use the check function to see which input errors are on the respective questionnaire

Errors are shown with a message text and displayed on the respective input lines



If you leave a screen, the data is saved automatically. If there are errors, it is shown on the next screen

If there are input errors, no results will be calculated at all


22

Input Questionnaire: User-Based Sizing Tables

3b

User sizing only considers concurrently active users

Most user sizings follow the standard pattern of low, medium, and high activity usersThink times between screen changes of 5 minutes, 30 seconds, and 10 secondsThe usual distribution is 60-70% low, 30% medium, 5-10% high


If in doubt, take medium activity users

The user sizing result is independent of the throughput sizing result

Choose Clear to delete entries

Input Questionnaire: Throughput Sizing Tables

3c

Entry fields / non-entry fields / mandatory input fieldsAll entry fields are whiteNon-entry fields are blue


Non-entry check boxes are greyMandatory input fields are marked with a star ( * )

Functions in the input tablesTray: Optional trays are folded (for example on EWM questionnaires)Mark lines Insert and delete lines or clear the entries for marked lines

23

Description of Key Input Fields and Functions in the Throughput Sizing Tables

Element: To identify the sizing element. On mouse over will display the long text in detailCan be business object, entity, scenario, …

A/P (A) (P) k

% chg / % dsp - changes & displays in %Number of changes and displays to an object in %. 1 change = 100%

Mon. = Months - Residence time in months

3c

A/P - (A)verage or (P)eakDetermines if a sizing is an average sizing or a peak sizing Peak time is not for disk sizing and geared at determining the absolute top CPU, memory, and I/O sizing

TI - Time intervalShows the chronological scope of the next field

(Y)ear: number of objects per yearPeak (P)eriod: Number of objects per defined time period(S)nap shot: Number of objects at anyone time

Objects - Number of objects Objects can be orders, projects, data records,…

Arch. - Archiving project plannedNo influence on sizing result

S.t. / E.t - Begin and end of processing timesWork day:

Can only be modified for entire questionnaireCannot be overwritten in input line

Peak period: Default can be changed for all empty lines Can be overwritten

ID Identification


ItemsSub objects such as line items, WBS elements, key figures, … per object

ID - Identification For checks and calculations of (multiple) average(s) and/or peak(s)

Short text - Text field for comments

Sizing Results

ResultsRelease, depending on level (SAP g (solution, software component)

CPUCategorySAPS

MemoryMB

Disk


CategoryMB

I/OCategoryI/Os per second

24

Details on Result Display / Result Level

Result levelsProjectSAP solution (default, according to SAP solution)Software component (separately installable software)Key capabilitySizing element (for analysis on element level)


g ( y )Line results + input (e.g. for multiple averages/peaks)Results, statistics, inputs (includes statistics for documentation)

Details on Result - With or Without Resource Requirements for a System Without Any Content (Offsets)

Per default, gross results are displayed for 'New system' (with minimal offsets). If you choose 'System extension‘, net results are displayed.On the project information questionnaire you can choose whether you want to see gross or net results on project solutions and software component resultsee gross or net results on project, solutions, and software component result levels. This choice is saved for the project.Additionally, you can switch between 'New system‘ and 'System extension‘ on the result page for project, solutions, and software component result levels. All other levels show net consumption


25

Sizing Results - In Categories, SAPS, MB, and I/Os per second

The results are displayed both in categories and SAPS / disk value / I/Os per secondAt the result levels project, SAP solution, software component, and key capability, the results are rounded to

100 SAPS1000 MB memory1000 MB disk

If your CPU sizing exceeds 30,000 SAPS or your disk sizing exceeds 1 TB or the I/O value is more than 12,000 I/Os per second, the Quick Sizer result should be handled with care.

Category Up to … SAPS Up to … MB Up to … I/Os per second

XS 2000 100,000 800


S 8000 250,000 3000

M 16,000 500,000 6000

L 24,000 1,000,000 10,000

XL 36,000 1,300,000 14,000

XXL Contact HW vendor or SAP

Sizing Results - With Charts and Details Options


26

Sizing Results - Statistics, e.g. on Disk

Overview of contributors for analysis (result level: results, statistics, inputs)

Table names


Table namesNumber of indexes per tableSize of each tableSize of all indexes per tableActually calculated disk size (depends on retention period)Disk size after one year (for growth estimation only)Information, if archiving object is available

Agenda


2. Sizing Basics





27

Sizing vs. Landscaping, Tuning and Configuration – Examples

LandscapingSizing resultStandard, advanced, expertOverall sizing KPIs

SAPS memory DB space disk I/O

2-tier, 3-tier, multi-tierHardware type, server type, memory, storage subsystem...SAPS, memory, DB space, disk I/O

By users, by throughputDistribution on servers/stacksDatabase/ABAP/Java/TREX/liveCache

TuningConfigurationRelease Enhancement package support package Parameter settings

subsystem...Production, test, QA, sandbox, ...Fail-over, Backup/recovery strategyData center layoutVirtualization (adaptive computing, cloud)Multiple components on one server


Release, Enhancement package, support packageProduct Availability Matrix

Processor speed – clock rate for response time or throughputDatabase, operating system, ...Interface design, Maintain server groupsAdministration settings, customizing

Parameter settingsE.g. Swap space, permanent space...

OLAP vs. OLTP parameter settingsResponse time vs. throughput tuningNumber of server nodes, threadsNumber of parallel jobsOptimal utilization of components

Validation of Sizing Assumptions

Load testing (Customer)SAP tools and methods (SAP)GoingLive Check by SAP support organization

Before go-liveCheck of Quick Sizer and other specifications with

Volume confirmationTesting of interfaces, backup strategy, ...Tuning of system parametersCheck of Quick Sizer and other specifications with

available hardwareEarly Watch Alert

Hardware utilization from SAPOSCOLContributors to load from statistical recordsDatabase and DB tables size and growthPart of Solution Manager

Miscellaneous (Customer)Vendor tools and methods (Partner) Hardware utilization information Do business figures fit to one another?

Tuning of system parametersResponse time tuningParallel job tuning

Optimal hardware utilization


Hardware utilization informationOperation system

Database monitors

Do business figures fit to one another?10,000 users per day vs. 100 business documents

Are all data neededAre there price/performance thresholds?Monitor the usage closely and adapt strategy, if necessary

28

Related Topics – Adaptive Computing

Quick Sizer Provides overall hardware forecasts to meet peak demands

Adaptive ComputingAdaptive ComputingAdaptive Computing dynamically assigns hardware resources

You can move a system for which a server is no longer providing the required performance to a more powerful serverDialog instances can be flexibly started on free servers to absorb spikes (or peaks) in workloadFlexible in unforeseen peak situations, avoids over-sizing and assures optimal capacity utilization

Does not replace sizing peaks still need to be met


Does not replace sizing peaks still need to be met

Agenda


2. Sizing Basics





29

General StatementsEffects of custom coding/3rd party often not predictable

A ”small” modification may have a large impact

Sizing Integrated Scenarios: Custom Coding/3rd Party/Cloud

MethodAnalysis: To what extend are bread-and-butter processes affected?

Small backend transactions without business impact may be omittedIf necessary, measure performance of test casesSAP delivers the toolset (CCMS, eCATT)For quality assurance, performance measurements ought to be done, anyhow

Use modular approach


You may use SAP standard sizing tools for the SAP-related sizingAdd the requirements of 3rd party, custom coding on topNote: some ISVs are required to provide standard sizing guidelines

3000 SAPS12 GB memory

800 SAPS3 GB memory




11,300 SAPS39 GB memory

ECC ServerSend orders to SCM via qRFCSend Advanced shipping

Interface Integration – Example of Complex Data Flow

User 1 User n

XI ServerSend ASN to ECCFrom ECC route ASN to SCM

http / XML

http / XML 4

5Send Advanced shipping notification (ASN) to SCM via XI

CRM ServerPerform ATP check in SCMS d h k d d li t ECC

qRFChttp / XML

qRFC

SCM ServerReceive ASN from XIR i d f ECC

2 35

User 1 User nUser 1 User n


1. CRM SCM (ATP check): sRFC (synchronous RFC, proprietary protocol)

2. CRM ECC (transfer unchecked delivery): qRFC (CRM middleware)

3. ECC SCM (transfer orders): qRFC (CIF, proprietary protocol)

4. XI ECC (transfer ASN): http / XML

5. ECC XI SCM (transfer Advanced Shipping Notification (ASN)): http / XML

Send unchecked delivery to ECC Receive orders from ECCSend orders to ECC

1sRFC

12345

30

Interface Integration – Load Caused by Communication

FileRFCBAPIsXML

Legacy SAP Business Suite

XMLSOAP ...

Procedure Size user and quantity structure individually for each systemAssume roughly 10% overhead for communication in sending and receiving system (IDoc, RFC)

Depends on size of the file each file has an administrative overheadError handling


Error handlingMonitoringTransactional behavior

Map to GUI transactionsCan be synchronous ( impact on response time!) or asynchronous

Integration to Business Intelligence

Data Extractor Business Warehouse

SAP Business Suite

ProcedureIn general contained in reporting, or may be added as "object display" loadBenefits through reduced reporting load


31

Test System

Dedicated for functional and non-functional testing

Possibly load testing

Tips: Sizing Development, Test, Sandbox and Q/A Systems

Development System

Developers on one or several technology stacks

Depends on application size power users

Sandbox SystemGoal: evaluation of functionality

Possibly load testingSize depends on usage

Size like production system or use minimum requirements

Q/A SystemSimilar to production system

Depends on application, size power users, consider memory requirementsRule of thumb: 20 developers on a 2-processor, dual core server


Goal: evaluation of functionalityA few users, some master dataSolution: minimum requirements http://service.sap.com/instguides

Similar to production systemSize like production systemIn 3-tier environment, usually only one app server

How to Size Initial Data Load

20,000,000 material master10,000,000 business partnersLegacy SAP Business

Suite

Initial data load versus master data managementIn general, even very high volumes compare little to document-type dataQuick Sizer only has limited master data sizing (SAP for Utilities, Banking, ...)SAP NetWeaver Master Data Management Sizing guideline

Procedure for initial upload


pRule of thumb for initial forecastPrecondition: ratio of DB:App = 1:51s for each object on one CPU

If the master data type is critical refer to SAP Notes, such as material master, where dedicated SAP notes show how to improve throughput for initial load

32

How To Make Sizing Measurements – A Bird’s Eye View

T t C

TextText

80/20 rule, simulate user behavior or backgroand processingRepresentative reproducible right to the point

Business Application, Scenario, Transaction

Test Case

Test System

Single-User Tests

Text– Text

Representative, reproducible, right to the point

Stable, no transports, no virtualization mechanismsCustomized, master data, system settings

Adherence to basic programming guidelines (e.g. accesses to DB, linear dependency,...) Scalability with the size of the business objects


Tests

Multi-User Load Tests

Scalability with the size of the business objectsScalability with the number of objects, Processing time, memory use, disk growth

Especially for Java-based applications, or high-throughputTo check locking mechanisms, stability, etc.Example: Standard Application Benchmarks ( SAPS)Hardware scalability, sizing verification

Load tests are also possible

Memory

Memory is usually dominated by user contexts methods different for ABAP and Java

ABAP

Which KPIs to Collect

CPU / Processing Power

CPU time per scenario or per interaction stepDetailed information

Database Disk Space and Disk I/ODisk space

ABAPUser context (maximum over test case)Memory in background jobs usually no issue

JavaUser session space, processing space, framework space

Frontend Network LoadTransferred amount of data

On applicationOn databaseBrowser rendering time, network time (only for response-time critical applications

consider synchronous traffic only)


pFine granular:

Number of inserts into tables during testDDIC information on table and index width (DB independent) multiply entries

CoarseDB snapshot before and after the test

Disk I/O: Monitor activity on ST04

Transferred amount of dataAverage number of KB sent and received per interaction stepCalculate bandwidth

Number of (synchronous) roundtrips between client and server

Impact on response time

33

Evaluating Measurement Results

Make sure you define at least three measurement points per scenario – the more, the better

Example: Scalability with the number of bidders – three tests

ApplicationDBTotal

0 1000

0.1200

0.1400

0.1600

0.1800

0.2000

U ti

me

Another set of tests would include, for example, 10 bids with 5, 50, and 150 line items, respectively


0.0000

0.0200

0.0400

0.0600

0.0800

0.1000

0 10 20 30 40 50 60 70No. of bidders

CPU

Example 2 – CPU time per interaction stepIn a 10-step scenario, each step takes 0.4s on average. Deductions

Assume think time of 30s between steps incl.

Examples of Sizing Calculations Based on Measurement Results

Example 1 – CPU time per scenarioYou measure 4s for a scenario of 10 interaction steps. Deductions

900 (3600s/4s) scenarios per hour, evenly

Example 3 – ThroughputA payroll run of 100,000 payroll periods takes 20

Assume think time of 30s between steps incl. system response time120 interaction steps per user per hour (3600s/30s) 12 scenarios at 100% utilization48 CPU s per user per hour (12*4s)75 users (3600/48) at 100% 49 at 65% (continuous work)

Example 4 – Lock on central objectA central object (e.g. material in goods issue) is

900 (3600s/4s) scenarios per hour, evenly distributed at 100% server/core/thread utilization590 (900*.65) at 65% utilization (SAP standard)590 scenarios (e.g. one per user per hour) net time, no think time included


A payroll run of 100,000 payroll periods takes 20 minutes on a single-processor, quad-core, dual-thread server. Deductions:

On a quad-processor, quad-core, dual-thread server with the same processors, the run should take 5 minutesOn a quad-processor, quad-core, dual-thread server with the same processors, 400,000 payroll periods could be processed in 20 minutes

A central object (e.g. material in goods issue) is locked for 200 ms Maximum throughput is 5 goods issues per second or 18,000 goods issues per hour. Poss. considerations if you need higher throughput

faster processoroptimization of codingchanging the business process ...

34

Production Sizings: General Procedure

ProcedureMonitor CPU utilization, table growth, and memory use

Relate it to a meaningful business entity, such as the number of concurrent users or the number of active projects

Re-Sizing Delta Sizing Upgrade Sizing

All projects All projects All projects

active projectsDifferent procedures depending on the puprpose

Re-sizing: Add the load from the additional users and projects causing the same load structure Delta sizing: treat like a new sizing and add calculated loadUpgrade sizing: determine additional requirements and add calculated load

Judge whether your current hardware is sufficient, or whether you may need to buy new hardware


All projectsSAP system monitorsGoal: Extend an existing system by load

E.g. by volume100 additional users who'll do the same as the current productive ones

All projectsSAP system monitorsGoal: Extend an existing system by functions

By different functions, e.g. you are live with CRM and want to add SCM

All projectsSAP system monitorsSAP NotesGoal: Upgrade SAP software

Standard Scenarios for Production Sizings – a Mixture of Different Approaches

Initial / delta sizing

Re-Sizing+ Quick Sizer + Quick Sizer + Load

+ Upgrade Increase in %

Upgrade sizing

Quick Sizer Current CurrentCurrent

ProjectProject

+ Upgrade Increase in %

Increase in %

+ Load Increase in %


Quick Sizer Project

Current Utilization

Current Utilization

Current Utilization

Upgrade and load increase (re-sizing)

New functions (delta sizing) and load increase (re-sizing)

New functions (delta sizing) and upgrade

Initial sizing

35

How to Monitor Current Resource Utilization

Available monitorsDisk Analysis DB02, DB monitor of vendor

(DB Performance Tables & Indexes)CPU Anal sis ST06 ST03N STAD ST03GCPU Analysis ST06, ST03N, STAD, ST03G

(Workload Analysis, Statistical Records, Global System Workload Analysis)User Analysis ST07, STAD, ST03G

(Application Monitor, Statistical Records)Memory Analysis SM04, STAD, GCLOG

(User List, Statistical Records)Front-End Network Load STAD, ST03N, ST03G, httplog

(Statistical Records, Workload Analysis)


As a rule, 20% of the processes cause 80% of the loadAnalyze

Growth rate of 20 largest tablesAverage and peak CPU loadAverage and peak memory utilization

Determine Current Utilization of KPIs

90%

100%

30%

40%

50%

60%

70%

80%


0%

10%

20%

30%

CPU Memory Disk Bandwidth

FreeConsumed

36

Upgrading Over Different Releases

In the standard case, simply apply several SAP NotesExample SAP ERP: SAP Notes 323263, 517085, 752532, 778774, 901070

To upgrade from release A to release CTo upgrade from release A to release CProcedure: Add the notes on resource requirementsBase release A release B release CExample: The SAP Notes say– A B +10%– B C + 5%Calculate: Utilization_A * 1.1 * 1.05 utilization_A * 1.16

CPU/Memory Resource Consumption relative to 4.6C1.80


Note: Ranges do not constitute a maximum

upper range

lower range

y p

1.00

1.10

1.20

1.30

1.40

1.50

1.60

1.70

4.6CSR1 47x110 47x200 ECC5.0 ECC6.0 4.6CSR1 47x110 47x200 ECC5.0 ECC6.0

CPU Memory

Range: approx. 15-35% Range: approx. 40-70%

Details on Upgrade Sizing

These effects may need to be consideredAdditional database usersIndirect effects due to buffer sizes, longer transactionsDifferent behavior of query optimizerChanges of the hardware and software configurationSome processes dominate resource consumption

Recommended procedure: Perform own measurements


37

Sizing in Print

Addresses sizing from different anglesMethodologyToolsTools VerificationCharacteristics of sizing projectsTechnical background

AuthorsSAP sizing experts

More informationwww sappress de or www sappress com


www.sappress.de or www.sappress.comISBN: 978-3-89842-796-8 (German)ISBN: 978-1-59229-116-8 (English)

SAP Developer Network www.sdn.sap.com: BlogsEfficient SAP Hardware Sizing: Quick Sizer

https://www.sdn.sap.com/irj/sdn/weblogs?blog=/pub/wlg/1826

Internet and Extranet Resources

Quick Sizer - Getting startedhttps://www.sdn.sap.com/irj/sdn/weblogs?blog=/pub/wlg/1961

Interpreting Quick Sizer Resultshttps://www.sdn.sap.com/irj/sdn/weblogs?blog=/pub/wlg/3102

SAP Service Marketplace http://service.sap.com/sizingGuidelines Process guidelinesGuidelines Solutions and platform


SAP Insider Article on using Quick Sizer for planning hardware expenditures (October 2009)

38

Virtual SAP TechEd :Extend your SAP TechEd Year Round

Best of SAP TechEd at Your fingertipsView sessions that you missedReplay and review sessions that you attendedattended

Quality SAP TechEd TrainingBest PracticesProduct Roadmaps

Learn at your own pace Gain Access to sessions recorded in 2006, 2007, 2008 and 2009* (*available December 2009)24/7 A li / ffli


24/7 Access online/offlineFlexible Course Syllabus

Volume LicensingSpecial Pricing for multiple subscribers

http://www.sdn.sap.com/irj/scn/virtualteched-allsessions

Further Information

SAP Public Web:SAP Developer Network (SDN): www.sdn.sap.comBusiness Process Expert (BPX) Community: www.bpx.sap.com

Related Workshops/Lectures at SAP TechEd 2009ALM108, Large SAP Systems: Performance Characteristics and Examples of Performance Challenges L1

SAP BusinessObjects Community (BOC): www.boc.sap.com


Examples of Performance Challenges, L1SOA203, General Performance Tuning with SAP NetWeaver Process Integration, L1

39

Th k Y !Thank You!

ContactFeedback

Please complete your session evaluation.p y

Be courteous — deposit your trash, and do not take the handouts for the following session.

40

No part of this publication may be reproduced or transmitted in any form or for any purpose without the express permission of SAP AG. The information contained herein may be changed without prior notice.

Some software products marketed by SAP AG and its distributors contain proprietary software components of other software vendors.

Microsoft, Windows, Excel, Outlook, and PowerPoint are registered trademarks of Microsoft Corporation.

IBM, DB2, DB2 Universal Database, System i, System i5, System p, System p5, System x, System z, System z10, System z9, z10, z9, iSeries, pSeries, xSeries, zSeries, eServer, z/VM, z/OS, i5/OS, S/390, OS/390, OS/400, AS/400, S/390 Parallel Enterprise Server, PowerVM, Power Architecture, POWER6+, POWER6, POWER5+,

Copyright 2009 SAP AGAll Rights Reserved

, , , , , , , , p , , , , , ,POWER5, POWER, OpenPower, PowerPC, BatchPipes, BladeCenter, System Storage, GPFS, HACMP, RETAIN, DB2 Connect, RACF, Redbooks, OS/2, Parallel Sysplex, MVS/ESA, AIX, Intelligent Miner, WebSphere, Netfinity, Tivoli and Informix are trademarks or registered trademarks of IBM Corporation.

Linux is the registered trademark of Linus Torvalds in the U.S. and other countries.

Adobe, the Adobe logo, Acrobat, PostScript, and Reader are either trademarks or registered trademarks of Adobe Systems Incorporated in the United States and/or other countries.

Oracle is a registered trademark of Oracle Corporation.

UNIX, X/Open, OSF/1, and Motif are registered trademarks of the Open Group.

Citrix, ICA, Program Neighborhood, MetaFrame, WinFrame, VideoFrame, and MultiWin are trademarks or registered trademarks of Citrix Systems, Inc.

HTML, XML, XHTML and W3C are trademarks or registered trademarks of W3C®, World Wide Web Consortium, Massachusetts Institute of Technology.

Java is a registered trademark of Sun Microsystems, Inc.

JavaScript is a registered trademark of Sun Microsystems, Inc., used under license for technology invented and implemented by Netscape.

SAP, R/3, SAP NetWeaver, Duet, PartnerEdge, ByDesign, SAP Business ByDesign, and other SAP products and services mentioned herein as well as their respective


logos are trademarks or registered trademarks of SAP AG in Germany and other countries.

Business Objects and the Business Objects logo, BusinessObjects, Crystal Reports, Crystal Decisions, Web Intelligence, Xcelsius, and other Business Objects products and services mentioned herein as well as their respective logos are trademarks or registered trademarks of Business Objects S.A. in the United States and in other countries. Business Objects is an SAP company.

All other product and service names mentioned are the trademarks of their respective companies. Data contained in this document serves informational purposes only. National product specifications may vary.

These materials are subject to change without notice. These materials are provided by SAP AG and its affiliated companies ("SAP Group") for informational purposes only, without representation or warranty of any kind, and SAP Group shall not be liable for errors or omissions with respect to the materials. The only warranties for SAP Group products and services are those that are set forth in the express warranty statements accompanying such products and services, if any. Nothing herein should be construed as constituting an additional warrant.

Documents

teched2009_ALM107