sap sizing rot 2011common1

© 2011 IBM Corporation

SAP Sizing ‘Daumenwerte’

Manfred Engelbart, SAP Solution Infrastructure Architect

Frankfurt, 24.3.2011


IBM SAP Alliance

2

IBM Rule of Thumbs for SAP Sizing

� Simple sizing approach based on best practice experiences

� Common Rule of Thumbs for p, i, (x)

� First guess for IT-landscape design

� Does not replace the formal SAP sizing procedure


IBM SAP Alliance

3

Best Practice Sizing Methodologies

� SAP Quicksizer (User and Quantity Based):

– CPU

– Capacity

– I/O-Performance

� IBM Rule of Thumbs:

– CPU

– Memory

– Capacity

– I/O-Performance

� Combination of Quicksizer and Rule of Thumbs

– CPU: RoT, User Based Sizing (SAP QS), Volume Based Sizing (SAP QS Throughput Sizing)

– Memory: RoT

– Disk Space: RoT plus SAP QS Results

– I/O Perf.: RoT or SAP QS Results


IBM SAP Alliance

4

CPU Sizing for frequently used SAP Components

- T-shirt Sizing (S/M/L)Solution

Manager

- SAP Quicksizer – Volume Based Sizing- based on Number and size of messages

PI (ehem. XI)

- SAP Quicksizer – User Based SizingEP (Portal)

- SAP Quicksizer – User Based Sizing (rough estimation)- Web-Reporting -> up to 80% of ABAP Load -> JAVA

BW

- SAP Quicksizer – User Based SizingCRM

- RoT- SAP Quicksizer - User Based Sizing- ISICC Sample Configs

ERP

Common Sizing MethodSAP

Comp.


IBM SAP Alliance

5

Rules of Thumb for SAPS (Unicode)

� RoT when not using the SAP QS

– 1 concurrent active ERP user = 10 SAPS

– 1 concurrent active CRM, BW user = 20 SAPS

– Complex application mix(incl. EP, PI, BW, ERP)= 22 SAPS (ISICC Sample Configs)

ASCII = Unicode minus 20% CPU capacity.

Named User= 100%,

Example: 100

Logged On User= 80% of

Named User Bsp: 80

Concurrent Active User= 65% of Logged On User,

Example: 80 * 0,65 = 52


IBM SAP Alliance

6

BS7 POWER SAPS/core overview http://www.sap.com/solutions/benchmark/sd2tier.epx

BS7 Unicode SAPS

0

500

1.000

1.500

2.000

2.500

3.000

3.500

8 16 16 32 8 8 32 32 48 64 32 128 256

Pow er5 Pow er5+ Pow er6 Pow er6+ Pow er6 Pow er7 Pow er7 Pow er7 Pow er7 Pow er7 Usparc

T2+

Sparc64

VII

Sparc64

VII

1.650 2.200 4.700 4.200 5.000 3.000 3.000 3.550 3.550 3.550 1.600 2.880 2.880

IBM IBM IBM IBM IBM IBM IBM IBM IBM IBM SUN/F SUN/F SUN/F

AIX 5.3

DB2

AIX 5.3

DB2

AIX 6.1,

DB2 9.0

AIX 6.1,

DB2 9.5

AIX 6.1,

DB2 9.5

AIX 6.1,

DB2 9.7

AIX 6.1,

DB2 9.7

AIX 6.1,

DB2 9.7

AIX 6.1,

DB2 9.7

AIX 6.1,

DB2 9.7

Solaris10,

Oracle

Solaris10,

Oracle

Solaris10,

Oracle

550Q p570 p 570 p 570 p 550 P 750 P 750 P 750 P 770 P 780 T5440 M9000 M9000

cores

processor

MHz


IBM SAP Alliance

7

BS7 x86 SAPS/core overview

http://www.sap.com/solutions/benchmark/sd2tier.epx

2.334

1.001

2.142

2.310 2.323

988

1.782 1.785

1.634

1.368

1.552

0

500

1.000

1.500

2.000

2.500

Intel

X5570

AMD

6176SE

Intel

X5670

Intel

X5670

Intel

X5670

AMD

6174

Intel

X7560

Intel

X7560

Intel

X7560

Intel

X7560

Intel

X7560

8 24 12 12 12 48 32 32 32 64 64

2.930 2.300 2.930 3.330 3.330 2.200 2.260 2.260 2.260 2.260 2.260

48 128 96 96 96 256 256 256 512 512 512

BL460c

G6

DL385

G7

BL460c

G6

DL380

G7

x3650

M3

BL685c

G7

DL580

G7

x3850

X5

RX600

S5

PQ

1800E

DL980

G7

HP HP HP HP IBM HP HP IBM Fujitsu Fujitsu HP

processor

cores

MHZ

GB RAM

model

2 Sockets 4 Sockets 8 Sockets


IBM SAP Alliance

8

Power6 vs. Power7 – SAPS/Core

3.145780 Turbo

2.781780 3.8 GHZ

2.725770 3.5 GHZ

2.348750 3.3 GHZ

2.670740 3.7 GHZ

2.327720 3.0 GHZPOWER7

2.2655.0 GHZ

2.140570 4.7 GHZ

2.1605.0 GHZ

1.8204.2 GHZ

1.520550 3.5 GHZ

1.8854.7 GHZ

1.700520 4.2 GHZPOWER6

SAPS/CoreModell/TaktrateProzessor Technologie


IBM SAP Alliance

9

SAP Sizing – CPU Rules of Thumb

� Per previous charts, we can conclude :

– 1 concurrent / active user requires 12 SAPS Backend (ECC)

– + 10 SAPS (EP, PI, BI) .

– = 22 SAPS SAP NW landscape

– 1 named SAP NW user requires 11 SAPS

� Now it‘s easy to map this to CPUs and Models

– tbd. (~2000-3000) SAPS/core for ECC 6.0 for POWER7 systems

– tbd. (~1000-2350) SAPS/core for ECC 6.0 for Intel/AMD based systems

2…3

2

1

CPUs min.

POWER

3…4

2... 3

1

CPUs min.

Intel/AMD

SmallestModel

Intel/AMD

5720260500

3432156300

114452100

SmallestModel

POWER

SAPS

NetWeaver

Concurrent

Users

Named

Users


IBM SAP Alliance

10

Old Memory RoT (ABAP)

� POWER5 at least 10 GB / Core

� POWER5+ at least 12 GB / Core



Remarks:

The values given are valid for 1 SAP instance. Each additional SAP instance will require at least 3 GB more memory

The values given are valid for a SAP typical load


IBM SAP Alliance

11

SAP Sizing – Some Memory Rules of Thumb

Server=

x+y SAPS

total capacity

QS SAPSresult = x

� Memory Recommendations

–6…8 GB per 1000 SAPS (Quick)sizer output

–assumes an ABAP and Java mix (80%:20%)

–Consequently, server capacity is MEMORY bound

–

� Above values are OK for a single SAP instance on a server/LPAR.

� Add a minimum of 2 GB for each instance in case you consolidate several SAP instances on a single server / partition.

� Consider some additional memory for virtualization features – see separate chart


IBM SAP Alliance

12

Virtual I/O Server Design – Basic recommendations

� Two Virtual I/O Server partitions are required for production load

– resilience against failure or misconfiguration– planned maintenance for VIO server

� In shared pools assign minimum 10% CPU power to the VIO Server+Clients

� Sample Virtual I/O Server partition configuration– 1 GB Memory (min=512MB max= 4GB)

– Uncapped Micropartitions with Capacity Entitlement = 0,5

– 2 Virtual CPUs, SMT enabled


IBM SAP Alliance

13

Memory RoT (Java)

� Java Basic Setup 3 GB

� Guideline for additional memory requiremends depending on SAPS value of Java application environments (first assumptions, subject to change).

Additional 3 GB Memory per:

– PI: approx. 600 SAPS (high memory footprint for large message sizes)

– BW: approx. 400 SAPS

– EP: approx. 800 SAPS


IBM SAP Alliance

14

Disk Space RoT

� Basis: 50 GB for OS, 100 GB for ‚empty‘ SAP system

� Use SAP QS results for capacity growth

– User based sizing: per year

– Throughput sizing: per retention period

� SAP QS results assumes single byte encoding, for Unicode apply approx. +50%

� temp Space

small installations (up to 1,000 SAPS) 50 GB

medium installations (up to 5,000 SAPS) 70 GB

large installations (greater than 5,000 SAPS) 150 GB


IBM SAP Alliance

15

Former Approach - I/O calculation

� Number of I/O per second required

– 0.4 I/O per second / SAPS (ERP, PI)

– 0.6 I/O per second / SAPS (BW)

– 0.2 I/O per second / SAPS, (assumption for SAP Components with low I/O requirements, e.g. EP, CRM, SolMan)

� I/O throughput

– 100 I/O per second (15k RPM drives)

– >1000 I/O per second (SSD drives, predominantly read)

� Example (15k RPM drives)

– 7,500 SAPS for ERP, PI approx. 3,000 I/O approx. 30 Drives

– 2,000 SAPS for BW approx. 1,200 I/O approx. 12 Drives

– 1,500 SAPS for EP approx. 300 I/O approx. 3 Drives


IBM SAP Alliance

16

SAP Sizing – Basic I/O Rules of Thumb

� 2,5 SAPS (DB+App-Sv) generate 1 I/O operation per second

� 1 concurrent SAP NW user generates ~9 I/Os per second

� A single 15k rpm disk is capable to support a maximum of 200 I/Os per second

– in other words: per 22 concurrent SAP users configure one disk drive

– resultingly, disk configuration is not capacity, but I/O driven

� Disk Controller

– RAID mechanisms have impact on aggregate I/O rates of storage subsystem.

• e.g., RAID 10 increases READ throughput (reads from 2 disks) butWRITE is reduced

� Adapters

– SCSI, FC, SATA, NAS, iSCSI

� Storage Sizing Guide available at:

– http://w3-03.ibm.com/support/techdocs/atsmastr.nsf/WebIndex/PRS3409


IBM SAP Alliance

17

SAP Sizing –DB-centric I/O Rules of Thumb (as of 2010) QS Approach

� 0,3 DB-Server SAPS generate 1 I/O operation per second

� New SAP I/O sizing approach for situations where DB-SAPS portion is well defined

– Previous RoT is considered to result in too high I/O estimates for newer SAP modules.

– Reason: more SAPS are consumed on App-Server side relative to DB

– SAP Quicksizer now explicitely shows SAPS split between the two instances.

� Variation of DB-SAPS : App-SAPS is significant for different SAP modules

– e.g., 1:3 for ERP = OK versus 1:15 for CRM = too high I/O load for DB-Server

– Their number is defined as own DB-Server SAPS requirement in SAP Quicksizer result section


IBM SAP Alliance

18

Danke für Ihre Aufmerksamkeit !

Fragen

sap sizing rot 2011common1

Documents