Hp COMPAQ PROLIANT 3000, COMPAQ PROLIANT 400, COMPAQ PROLIANT 4500, COMPAQ PROLIANT 1200, COMPAQ PROLIANT 5500 SAP R/3 on the Compaq ProLiant Platform

WHITE PAPER

.

[July 1995] Compaq Computer Corporation

CONTENTS

Introduction ................... 1

The Compaq-SAP

Partnership.................... 3

The International Compaq SAP

Competence Center ............. 3

Regional Compaq SAP

Competence Centers ........... 3

Certification of SAP

R/3 ................................. 4

Process .............................. 4

Currently Certified Platforms 5 SAP Logo Partner Program .. 5

SAP R/3 System

Architecture ................... 7

Overview............................. 7

The SAP R/3

Application Services ............. 8

System Configuration .... 9

System Performance and

Benchmarking ..................... 9

System Sizing

Methodology ...................... 13

Disk Configuration ............. 17

Networking ........................ 18

General Hardware

Configuration Issues .......... 20

Operating System .............. 21

Example Systems .............. 23

System Installation

Guidelines ......................... 26

Distributed Systems ........... 30

Upgrading SAP

R/3 Systems ...................... 32

Backup and Restore of

R/3 Systems ................. 33

Operating System Backup

and Restore ...................... 33

Executables Backup

and Restore ...................... 33

Database Backup

and Restore ...................... 34

Technical Support .............. 37

SAP R/3 Support ............... 37

Compaq Insight Manager ... 39

Training Requirements. 39

Introduction ....................... 39

Need for Training within

the Compaq Community ..... 39

Application Training ........... 40

Basis Training ................... 40

Installation Certification ..... 40

Appendix: Example

Production System ....... 42

.

SAP R/3 on the Compaq ProLiant Platform

.

Introduction

.

SAP R/3 is an integrated business application and is the successor to the mainframe-

.

based R/2 product. R/3 was first available in July 1992 but was confined to the UNIX

.

environment. In October 1993, SAP announced the future availability of SAP R/3 on

.

the Microsoft Windows NT platform and the product was launched at the CeBIT fair in

.

Hannover in March 1994.

.

This document provides a technical overview of the Compaq SAP R/3 platform, with a

.

targeted audience of Compaq pre-sales and support personnel, SAP partner

.

organizations, and technical departments in customer organizations. The document is

.

not intended to be an installation guide for the system, but should provide all the

.

technical information necessary to plan an installation effectively and to supplement

.

the skills of a Certified R/3 Installer.

.

Complementary documents are the iXOS certification report and the SAP R/3

.

installation guide. A benchmarking environment has now been established at the

.

Compaq Competence Centers in Munich and Houston. As well as providing

.

information for marketing and system sizing, this environment will be used for

.

generating technical documentation (optimal disk layouts, backup strategies, tuning,

.

and so on), performing Proof of Concepts for customers, and integration testing.

.

The mailing address of the Compaq SAP Competence Center EMEA is:

.

cpqsap@bangate.compaq.com

230A/1095 1

WHITE PAPER (cont’d)

.

NOTICE

.

The information in this publication is subject to change without notice.

.

COMPAQ COMPUTER CORPORATION SHALL NOT BE LIABLE FOR TECHNICAL OR

.

EDITORIAL ERRORS OR OMISSIONS CONTAINED HEREIN, NOR FOR INCIDENTAL

.

OR CONSEQUENTIAL DAMAGES RESULTING FROM THE FURNISHING,

.

PERFORMANCE, OR USE OF THIS MATERIAL.

.

This publication contains information protected by copyright. Except for internal use

.

distribution, no part of this publication may be photocopied or reproduced in any form without

.

prior written consent from Compaq Computer Corporation.

.

This publication does not constitute an endorsement of the product or products that were tested.

.

The configuration or configurations tested or described may or may not be the only available

.

solution. This test is not a determination of product quality or correctness, nor does it ensure

.

compliance with any federal, state or local requirements. Compaq does not warrant products

.

other than its own strictly as stated in Compaq product warranties.

.

Product names mentioned herein may be trademarks and/or registered trademarks of their

.

respective companies.

.

Compaq, Systempro, Compaq Insight Manager, Fastart, registered U.S. Patent and Trademark

.

Office.

.

ProSignia, ProLiant and QVision are trademarks of the Compaq Computer Corporation.

.

Microsoft, Windows, Windows NT, Windows NT Advanced Server, SQL Server for Windows

.

NT are trademarks and/or registered trademarks of Microsoft Corporation.

.

SAP R/3 on the Compaq ProLiant Platform

.

First Edition (July 1995)

.

Document Number 230A/1095

.

230A/1095

2

WHITE PAPER (cont’d)

.

THE COMPAQ-SAP PARTNERSHIP

.

The International Compaq SAP Competence Center

.

Compaq and SAP have established an International Competence Center (ICC) at the SAP HQ

.

in Walldorf, Germany. The ICC has dedicated resources as part of the SAP support

.

infrastructure and will be active in joint engineering projects.

.

Key activities carried out at the ICC are:

.

• SAP second-level hotline support

.

• Joint engineering development projects with SAP

.

• Post-sales critical customer support

.

• Provision of early watch services

.

• System sizing

.

• Development of support tools

.

• Customer briefings at Walldorf

.

• Marketing

.

• Event support

.

Regional Compaq SAP Competence Centers

.

In addition to the International Compaq Competence Center, Compaq has established regional

.

Competence Centers:

.

Regional Responsibility Location

.

Europe, Middle East, and Africa Munich, Germany

.

North America Houston, USA

.

Japan Tokyo, Japan

.

Asia Pacific Singapore

.

Key activities carried out at these centers are:

.

• Pre-sales support to subsidiaries and partners

.

• Pre-sales critical customer support

.

• Major account briefings

.

• Platform certification

.

• Benchmarking

.

• System sizing methodology and sizing service

.

• Platform specific engineering projects, such as developing backup/restore solutions

.

• Development of Lotus Notes-based project management tools

.

• Publication of White Papers

.

• Marketing

.

• Event support

.

230A/1095

3

WHITE PAPER (cont’d)

.

CERTIFICATION OF SAP R/3

.

Process

.

SAP produces a single version of R/3 for Windows NT. This version runs on all Intel-based

.

Windows NT hardware platforms. No additional software build is necessary to run R/3 on Intel

.

Windows NT platform, unlike UNIX, where each vendor platform requires a separate

.

compilation of the R/3 executables. This is not the case for Windows NT systems using other

.

processor architectures. However, for each hardware platform R/3 is released on the same date

.

and has the same functionality.

.

SAP does not support all Windows NT hardware platforms (of which there are hundreds). To

.

test the ability of a particular hardware platform to support R/3, SAP introduced a certification

.

procedure. This certification is carried out by a partly owned partner of SAP, iXOS Software

.

GmbH, who established the “Competence Center for SAP R/3 on Windows NT” (R/3 NTC).

.

The certification procedure checks that each platform is technically sound and performs well

.

with R/3, as well as having the necessary service and support infrastructure in place to support

.

R/3 implementations.

.

Each platform certification relates to the following:

.

• A major R/3 version (for example, 2.1 or 2.2)

.

• A Windows NT version (for example, 3.1 or 3.5)

.

• A database version (for example, Oracle 7.0, Oracle 7.1, Adabas 3.12, SQL Server 6.0)

.

• A hardware architecture (for example, Compaq ProLiant)

.

As an example, the Compaq Proliant has been certified for R/3 release 2.2, Oracle release 7.1

.

and Windows NT 3.5. Compaq ProLiant is certified as a system family. This means that

.

product line enhancements like ProLiant 4500 do not need separate certification.

.

There is only one common release schedule for all supported platforms. This means that the

.

Windows NT version of R/3 ships at the same time as the UNIX version. The certification

.

process for all Windows NT platforms can sometimes take up to three or four weeks. A

.

customer may use a non-certified version of the system. However SAP may not give timely

.

support for such a non-certified system so that certainly for productive systems only certified

.

versions should be used.

.

Only number releases of R/3 require certification, for example, 2.1, 2.2, or 3.0. Alphabetical

.

releases (correctional stages) such as 2.2b, 2.2c or 2.2d do not require additional certification. A

.

customer can go productive with these releases as soon as they are made available by SAP.

.

230A/1095

4

WHITE PAPER (cont’d)

.

Currently Certified Platforms

.

The SAP release schedule for R/3 on Windows NT version is the same as for UNIX versions:

.

R/3 NT Database Availability

.

Release 2.1 J 3.5 Oracle 7.1.3.3.6 today

.

Release 2.2 E 3.5 ADABAS-6.1.1.13 today

.

Release 2.2 F 3.51

.

Release 3.0B 3.51

.

In the Compaq ProLiant Family of Servers, the EMEA Competence Center recommends that

.

the ProLiant 1500 not be offered for productive environments due to the limited scalability in

.

memory (maximum 208 Megabytes) and system processors. With a clear notification of these

.

limitations, the system can be offered to customers as a test or training system.

.

SAP Logo Partner Program

.

The SAP R/3 Logo Partner program encourages other companies to provide marketing,

.

consulting, and service activities related to SAP R/3. Many different types of companies are

.

Logo Partners, for example, large consulting companies or small service providers. The R/3

.

Logo Partner agreements are valid for two years. SAP reviews these agreements regularly to

.

ensure quality of services and high customer satisfaction.

.

As part of the Logo Partner program, SAP offers the following:

.

• Free SAP R/3 demo licenses

.

• Dedicated second-level support

.

• Special education programs

.

• Common marketing events

.

• Regular SAP R/3 partner meetings

.

• Access to SAP internal systems

.

• Cooperation in development projects

.

• Project support through the SAP partner services

.

In return, the Logo Partners have an obligation to build up expertise in multiple areas (Business

.

Process Re-engineering - BPR, R/3 base system, application modules, and so on) to assist

.

customers implement and deploy SAP R/3 solutions. The Logo Partners do not sell SAP R/3.

.

They cooperate with the SAP sales force.

.

3.51

Oracle 7.1.3.3.6 ADABAS-D 6.1

SQL Server 6.0 Oracle 7.1 ADABAS-D 6.1 SQL Server 6.0

today in the process of certification

in the process of certification planned for 4Q/95

230A/1095

5

WHITE PAPER (cont’d)

.

Global SAP R/3 Logo Partners operate in all regions where SAP operates and have the

.

infrastructure required to support international R/3 customers. The following list contains

.

some of these Logo Partners:

.

• Andersen Consulting

.

• BSO Origin

.

• CAP GEMINI

.

• Coopers & Lybrand

.

• Deloitte & Touche Tohmatsu

.

• Ernst & Young

.

• I • C • S

.

• KPMG

.

• Price Waterhouse

.

• EDS

.

National SAP R/3 Logo Partners operate predominantly within one country.

.

230A/1095

6

WHITE PAPER (cont’d)

.

SAP R/3 SYSTEM ARCHITECTURE

.

Overview

.

R/3 supports a three-level client/server architecture. Depending on the configuration of the

.

complete R/3 system, the three major software components described below could all run on

.

one physical machine or on several physical machines:

.

• All database transactions are handled by the Database Server. Currently, there can only be

.

one physical machine running the Database Server software component.

.

• The Application Server (AS) processes the input within the SAP R/3 work processes.

.

Because all parts of the R/3 software except the kernel are written in ABAP/4, an

.

interpretative 4GL developed by SAP, the application server is very processor intensive.

.

To increase processing power, several physical machines can be added to the R/3 system

.

that run the Application Server software component.

.

• The Presentation Server (PS) supplies the user interface to the R/3 application. These are

.

primarily Windows 3.xx based, but Windows NT, OSF/Motif (X11), OS/2 Presentation

.

Manager, and Macintosh are also supported. Nearly 90 percent of the R/3 front-ends are

.

Windows based. The presentation server can also provide graphical analysis of application

.

data.

.

If the Database Server and Application Server are running on the same physical machine, the

.

system is referred to as a Central System. It is possible to separate the database and application

.

server across different systems. In fact, a single database server can support several application

.

servers. This is known as a Distributed System (DS). In all cases, the network interconnection

.

between the R/3 components is made over the TCP/IP protocol.

.

The performance of an SAP system is very much dependent upon the Application Server

.

component. Typically, there is a 4:1 ratio between the work done by the Application Servers

.

and that done in the Database Server. In a distributed environment, if all servers were

.

similarly configured, there would be one Database Server and four Application Servers.

.

230A/1095

7

WHITE PAPER (cont’d)

.

The SAP R/3 Application Services

.

The SAP R/3 system is a "true" client/server system. This means that a user logged on to the

.

SAP R/3 system implicitly uses processes from all the servers in the network. The specific

.

process available on the server and the extent of the process depends on the configuration of

.

the SAP R/3 system.

.

Process Description

.

Dialog Process Interactive transactions of the SAP R/3 system.

.

Batch Processes The SAP R/3 system contains a complete job-scheduling system.

.

Update Processes All database changes, adding a sales order for example, are done

.

Enqueue

.

Processes

.

Spool Processes The SAP R/3 system contains a comprehensive spooling system.

.

Message Server Manages communication between the work process, within a server

.

When R/3 is started each server can run a different group of workprocesses, depending upon its

.

role. In a central server configuration all processes run on the same server. In a distributed

.

environment processes can be distributed between servers, although in most cases all processes

.

but the dialog process run on the central server.

.

This service is also used to put long running queries into background. Batch processes are basically the same entity as a Dialog Process, but do not accept user input. In fact, the system can be configured to convert Dialog Processes between batch work and on-line work during different periods of the day.

asynchronously to the dialog process. SAP implements its own database lock manager, known as the

Enqueue Process. Only two Enqueue processes can be started and must be run on the same server in a Distributed System.

To print spool orders, Spool Processes are required.

and between separate servers. Only one message server exists per installation of R/3. With R/3 Version 3.0, the message server also handles the load balancing of user logons across application servers, allocating new logons to the least active system.

230A/1095

8

WHITE PAPER (cont’d)

.

SYSTEM CONFIGURATION

.

R/3 is a very intensive application, requiring considerable memory, processor capacity, and

.

disk resources. This section presents some guidelines in configuring Compaq ProLiant systems

.

to host the R/3 application. Example system configurations are provided for minimal and

.

development systems, together with advice on the disk layout for larger production systems.

.

These are not intended to be specific guidelines for the sizing of productive systems. More

.

details can be obtained from the Competence Centers.

.

System Performance and Benchmarking

.

Benchmarking Methodology

.

SAP has defined a standard set of benchmarks for the different application modules (SD, FI,

.

MM, PP, HR). These benchmarks are available to all hardware partners to enable them to

.

provide competitive performance analysis, together with system capacity planning. The

.

benchmark suite also provides a very valuable means to place a substantial load on a system

.

during testing of new hardware and software components.

.

Benchmarking is an open process in which all of the SAP hardware partners participate. The

.

benchmarks are controlled by SAP but run by the individual companies. This process is

.

overseen by the “SAP Benchmarking Council,” an informal grouping of SAP and the hardware

.

partners. This council monitors the benchmarking activity and provides standards for reporting

.

the results, which must be verified by SAP prior to publication.

.

The benchmark consists of a number of script files which simulate the actions of a typical user

.

in each of the modules and a predefined SAP client database which contains simulated

.

company data against which the benchmark is run.

.

After a standard installation of the R/3 system and import of the predefined company data, the

.

script files are executed on a benchmark driver system. Currently, this system must be a UNIX

.

workstation. However, a Windows NT version is currently under development. Each script file

.

simulates the entries made by a user in the SAPGUI frontend software, executing a variety of

.

SAP transactions consisting of one or more screen changes or dialogue steps (DS). This

.

process employs the playback mechanism in the SAPGUI application allowing keystrokes to be

.

saved and replayed against the frontend software. This means that for each benchmark user

.

simulated a separate SAPGUI process must be run.

.

After each dialog step there is a 10-second pause called the think time. SAP has experienced

.

that an average user performs a dialog step every 30 seconds. Therefore, the load imposed by a

.

single script file represents roughly three active users (when taking the system response time

.

into account, this ratio is more accurately specified as 2.67, and for planning purposes we

.

generally apply a factor of 2.5).

.

230A/1095

9

WHITE PAPER (cont’d)

.

Once a script file performs all its SAP transactions, it repeats these steps in a loop. The

.

repetition time may be defined when commencing the benchmark. Typically, a benchmark

.

simulating a specified number of users (and therefore the same number of SAPGUI processes)

.

must be run for a period of one hour to generate statistically significant results.

.

After each run, performance data is retrieved from the SAP system and collected on the

.

benchmark driver system. Then the next run is performed with an increased number of

.

users. The benchmark result is defined as the number of benchmark users supported by the

.

system in the given application module, with an average dialog process response time of

.

two seconds.

.

This figure should be viewed as a measure of the absolute performance capability and is

.

primarily intended to enable comparisons between different hardware platforms. This figure

.

should never be used as part of the capacity planning process. For system sizing, a further

.

criterion is applied:

.

The number of benchmark users supported by the system in the given application module,

.

with an average dialog process response time of less than two seconds and processor

.

utilization on the database server of no more than 60 percent.

.

This extra constraint allows for the fact that systems experience peak loading and that

.

background processes, such as print spooling and batch jobs, are always present. On a well-

.

balanced hardware architecture, such as the Compaq ProLiant, two-second response times

.

only occur with the system processor utilization at 100 percent. Generally, the response

.

time experienced in a system at 60-percent utilization is below one second.

.

The response time definition is based upon the dialog process. A real user experiences a longer

.

response time due to delays imposed by the network and performance of the workstation used.

.

If both these components are correctly specified, the increased time should not be a problem.

.

Although benchmarks are available for the majority of the SAP modules, the figure most often

.

communicated is the number of SD benchmark users supported. This has replaced the earlier

.

use of FI users. The SD application is one of the most complex in the R/3 suite and provides

.

the best single representation of a mixed group of users accessing different modules.

.

Publication of Results

.

The result of the benchmarks may be communicated in two different ways:

.

1. Number of benchmark users in a given module.

.

2. The number of SAPs supported by the system.

.

Method 1 is only valid when communicated together with a mean response time and mean

.

processor utilization.

.

Method 2 is a measure of total system throughput and enables results from different SAP

.

module benchmarks to be compared. SAP has defined a conversion factor per benchmark to

.

generate this common throughput-based result. The prime conversion factor is based on the

.

benchmark module SD, and 100 SAPs are defined as: “Updating 2,000 business operations per

.

hour in the SAP application benchmark SD. In this time, 6,000 dialog steps are executed."

.

While it is valuable, this figure is rarely published. Generally, the simpler user figure is given.

.

230A/1095

10

WHITE PAPER (cont’d)

.

Interpretation

.

As mentioned above, the intent of benchmarks is to provide a means of comparing systems. It

.

is not possible to guarantee a real user response time using the benchmark suite, because the

.

performance impact of customizing, database size, batch loads, and specific system

.

configuration cannot be included.

.

With two given results you can establish a comparison as follows:

.

Example 1 - SAPs

.

Interpretation: System B is able to handle 50 percent more dialogue steps than System A in the

.

same amount of time.

.

It does not mean:

.

Example 2 - SD Users

.

This result provides very little information.

.

This provides at least a part of the information needed. System B is handling 30 percent more

.

users with a 50 percent longer mean response time per dialogue step.

.

This is the complete information and shows that System B is definitely more powerful than

.

System A, because B can sustain the response time for even more users or while executing

.

background batch jobs. System A shows severe response time degradation when adding

.

additional load.

.

System A: 200 SAPs System B: 300 SAPs

System B is faster than System A; System B could have more but slower processors than System A, resulting in a poorer mean response time but higher throughput.

System A: 300 SD benchmark users System B: 400 SD benchmark users

System A: 300 SD benchmark users, 1 second MRT/DS System B: 400 SD benchmark users, 1.5 second MRT/DS

System A: 300 SD benchmark users, 1 second MRT/DS, 95-percent mean processor utilization System B: 400 SD benchmark users, 1.5 seconds MRT/DS, 65-percent mean processor utilization

230A/1095

11

WHITE PAPER (cont’d)

.

Central System Performance

.

The application scales very well with the number of processors, up to and including the

.

maximum of four:

.

Processor Utilization 60% 100% 60% 100%

.

1 x 100-MHz Pentium Processor 23 38 63 63

.

2 x 100-MHz Pentium Processors 43 68 133 175

.

4 x 100-MHz Pentium Processors 83 113 175 268

.

4 x 133-MHz Pentium Processors 110 163

.

NOTE: With the 133-MHz boards we saw an excellent performance improvement, in the range

.

of 30 to 40 percent. The reason is not only the higher clock speed but also the improved cache

.

design.

.

SAP recommends designing systems for average processor loads that are close to 60 percent.

.

However, the 100-percent figure indicates the potential for handling peak loading during

.

exceptional circumstances. The 60-percent figure must be used for the system sizing process.

.

The maximum number of users supported by a given configuration will be reduced if batch

.

jobs are being run at the same time as user access. Also, only a properly designed system can

.

achieve high throughput. When there is a disk or memory bottleneck, additional processors

.

never result in more throughput.

.

Distributed System Performance

.

Once user requirements exceed the performance of a central server configuration, it is

.

necessary to split the database and application components across separate systems. The

.

performance of such a system is very dependent upon the components used at all levels of the

.

system. However, the database server always provides the ultimate performance bottleneck.

.

Recent benchmarks, utilizing the Oracle database, have demonstrated the following

.

performance capabilities of the Compaq ProLiant platform:

.

Maximum Number of

Active Users in SD

Module

Maximum Number of

Active Users in FI

Module

230A/1095

12

WHITE PAPER (cont’d)

.

Processor Utilization 60% 80%

.

Database Server

.

4 x 100-MHz Pentium 450 550

.

Application Servers

.

2 x 100-MHz Pentium 63

.

4 x 100-MHz Pentium 116

.

These results support the observation that there is approximately a 4:1 ratio between the

.

workload on an application server and that on a database server. In this case, with all

.

components at

.

80-percent utilization, the database server can support 4.75 times the number of users that can

.

be supported on an identically configured application server. The latest results can be

.

retrieved from the Compaq SAP Competence Center by mailing to

.

cpqsap@bangate.compaq.com.

.

System Sizing Methodology

.

To perform a successful sizing, the customer must provide information concerning the planned

.

installation. A questionnaire for gathering this information is available from your local

.

subsidiary, Regional Competence Center, or the International Competence Center.

.

The required information falls into two categories:

.

• First, we must know the peak number of users accessing the system in each of the

.

different SAP modules, for example, Financials, Materials Management, Sales and Distribution, and so on. These must be further divided into active and in-active but logged-on users. This differentiation is important. Even if a logged-on user is not working, and therefore not creating any processor load, the system still reserves memory for that connection. The definition of an active user is one who submits a screen change in SAPGUI every 30 seconds, viewed by SAP as the average work rate experienced on customer sites. This information is used to size the number and type of processors and the amount of memory required.

• Secondly, information is needed about the transactions and data volumes on the system.

This information is specific to each of the SAP modules and requires a detailed understanding of the business processes being implemented. This information is used to size the database.

We very strongly recommend that customer sizing information be gathered by a knowledgeable SAP consultant. In particular, small errors in the disk sizing questionnaire might lead to large errors in the amount of disk space predicted for storing the database tables.

If the sizing questionnaire is correctly filled in, sizing an R/3 system should not take more than a few hours. However, these questionnaires are often incomplete or produce unrealistic configurations due to errors in answering the questions. From our experience we suggest allowing a minimum of 2 weeks from the point of submitting the questionnaire to completion of a system proposal. In fact, for larger systems this might take even longer.

13

Maximum Number of Active SD Users

Supported per server

230A/1095