SAP CRM Benchmark

pages

English

Documents

Écrit par
Morten Loderup

Publié par
Misheg

Lire

Le téléchargement nécessite un accès à la bibliothèque YouScribe Tout savoir sur nos offres

pages

English

Documents

Lire

Le téléchargement nécessite un accès à la bibliothèque YouScribe Tout savoir sur nos offres

Publié par

Misheg

Nombre de lectures

Langue

English

SAP CRM Benchmark
on
Dual-Core Dell Hardware

Morten Loderup
Dell SAP Competence Center
28 August, 2006
Dell Inc.

Contents

Executive Summary…………………………………………………………….……3

SAP CRM Software – a brief introduction…………………………………….…….4

CRM Project…………………………………………………………………….…...5
Hardware certification and benchmark………………………………….…...5
Hardware and software used…………………………………………….…...5
Configuration Steps……………………………………………………….….6

Benchmark findings……………………………………………………………….….8
Number of Instances………………………………………………………….8
Number of Dialog work-processes………………………………………….10
Sequential Memory setting………………………………………………….11

Conclusion…………………………………………………………………………...12

Appendix – Vocabulary……………………………………………………………...13

2
This white paper discusses findings from performing the SAP Customer Relationship
Management (CRM) benchmark and provides considerations for implementing SAP’s
CRM software in a Dell hardware environment.

Executive Summary
SAP provides Customer Relationship Management (CRM) software for the following
business areas: Marketing, Sales, Service, E-commerce, Interaction center operations and
management, and Channel Management. According to the research company AMR
Research, SAP is growing faster then Siebel (largest CRM vendor) in terms of sales
growth (“SAP preps for fresh wave of CRM activity,” Computer Business Review,
10/6/05, http://www ...

Voir

Publié par

Misheg

Langue

English

Sapois Institutions de l'Union européenne Linternaute.com Carte MS Configuration (informatique) Diagramme de Bode

SAP CRM Benchmark noDual-Core Dell Hardware Morten Loderup Dell SAP Competence Center 28 August, 2006 Dell Inc.

Contents Executive Summary.3 SAP CRM Software a brief introduction..4 CRM Project....5 Hardware certification and benchmark....5 Hardware and software used....5 Configuration Steps..6 Benchmark findings..8 Number of Instances.8 Number of Dialog work-processes.10 Sequential Memory setting.11 Conclusion...12 Appendix Vocabulary...13 2

This white paper discusses findings from performing the SAP Customer Relationship Management (CRM) benchmark and provides considerations for implementing SAPs CRM software in a Dell hardware environment. Executive Summary SAP provides Customer Relationship Management (CRM) software for the following business areas: Marketing, Sales, Service, E-commerce, Interaction center operations and management, and Channel Management. According to the research company AMR Research, SAP is growing faster then Siebel (largest CRM vendor) in terms of sales growth (SAP preps for fresh wave of CRM activity, Computer Business Review, 10/6/05, http://www.cbronline.com/article_news.asp?guid=0BCB0BB2-9685-4CD0-BB35-8B396F01D7C0 ). Dell SAP Competence Center performed the SAP Interaction Center (IC) standard mySAP CRM 2005 application benchmark (hereafter referred to as the CRM benchmark) to assist the Dell SAP custome rcommunity to gain a better understanding of the hardware requirements when implementing the SAP CRM solution. The CRM benchmark solution consisted of the following hardware and software: a Dell PowerEdge 6850 server with a Dell/EMC CX 600 storage array, Microsoft Windows 2003 Enterprise Edition operating system (64-bit), Microsoft SQL 2000 database, and SAP Customer Relationship Management software version 5.0 (SP 41). The benchmark testing gives insight into the variables that affect performance of SAP CRM software. Among the many test runs completed, the following variables have a significant effect on the performance of SAPs CRM software in a Dell hardware environment: Number of Instances Number of dialog work processes Sequential Memory setting As a result of Dell performing this benchmark, SAP CRM customers now have a better understanding of the impact of SAP Instances, dialog work processes, and sequential memory settings when implementing SAPs CRM solution in a Dell hardware environment. Moreover, Dell has become the first vendor in the history of SAP to offer a CRM benchmark. 3

SAP CRM Software a brief introduction SAP Service Oriented Architecture is SAPs paln for a business process-driven approach to enhance competitive advantage, dynamically adjust to changes, and ensure constant productivity. SAP CRM is built on SAP NetWeaver technology and Service Oriented Architecture. CRM, SAPs Customer Reltaionship Management software, enables businesses to become more integrated and more efficient, with instant access to critical real-time information. Thus enabling small, mid-size, and enterprise customers to make competitive edge decisions. SAP provides CRM software for the following business areas: Marketing, Sales, Service, E-commerce, Interaction center operations and management, and Channel Management. In addition to SAP, there are other serious CRM contenders in the market place. SAPs major competitors include Siebel (now a part of Oracle), Microsoft, Salesforce.com, RightNow, SalesNet, NetSuite, and others. Of all these competitors, Siebel is the largest with the most number of CRM seats. According to AMR Research, SAP has 160,000 live CRM users compared to Siebels 3,200,000. However, SAP is growing faster then Siebel from a revenue perspective. (SAP preps for fresh wave of CRM activity, Computer Business Review, 10/6/05, http://www.cbronline.com/article_news.asp?guid=0BCB0BB2-9685-4CD0-BB35-8B396F01D7C0 ). There are several reasons why SAPs CRM is becoming so popular. Tight integration with all components of the customer relationship cycle, a sound Service-Oriented architecture, and a robust, time-tested SAP foundation seem to be favored by customer choice. Integration with telephony and email, Internet, mobile clients, handhelds, portal, and existing SAP systems (like Enterprise Core Component (ECC), Business Warehouse (BW), and Supply Chain Management (SCM)) enhancing the capabilities of CRM gives additional value-add. 4

CRM Project The motivating factors and goals for running the CRM benchmark were to assist the Dell SAP customer community to gain a better understanding of the tuning requirements when implementing a SAP CRM solution. Moreover, Dell has become the first vendor in the history of SAP to offer a CRM benchmark. By so doing, Dell demonstrated it is willing to make the investment to help ensure customer peace of mind for a CRM landscape, and it has the in-house expertise to do so. Hardware certification and benchmark In order for customers to better understand the hardware requirements for running a particular SAP application, SAP provides certification and benchmarking programs. Of all the SAP certification programs available, the SD-benchmark is most common. After a benchmark and certification is submitted to SAP, the concluding information becomes public domain for customers to access. A customer-focused goal of benchmarking is to create a representative environment of what the customer might expect running the solution under test. To create an optimal environment where a large number of users access the hardware with low response times, a balance between hardware, software and tuning parameters must exist. Hardware and software used The hardware used for the CRM benchmark included a Dell PowerEdge 6850 server with Dual-Core Intel Processors, connected to a Dell/EMC Storage Area Network (SAN). Software used to configure the solution included SAP CRM 5.0 running on Microsoft Windows 2003 64-bit Operating System, and with a Microsoft SQL2000 database. PE6850 CX600 Storage Array PDreollc eSsesrovre r P4 oInwteerl EXdegoen 638.50 0 GHz DC Memory 32 GB SAN Dell/EMC CX600, QLogic 2460 HBA (2), Brocade 3800 Fibre Channel Switch (2) LUN RAID RAID 10 SAP Software CRM 5.0 (SP Level 41) Operating System Microsoft Windows 2003 EE 64-bit Database Microsoft SQL2000 (SP 1) Benchmark Client PowerEdge PE2500, Pentium III, 993Mhz, 4GB memory, MS W2K3 EE x86_32-bit Table 1: Hardware and software configuration 5

Configuration steps The configuration steps for setting up the Dell Storage Area Network and SAP CRM software consisted of the following: 1) Configure Storage Area Network (SAN) Create LUNS (RAID10) Update FLARE code (if necessary) Zone Brocade switches 2) Configure Application/Database Server update BIOS and Firmware, install Microsoft W2K3 x86_64-bit, configure HBAs, install Dell|EMC PowerPath, NaviAgent, and SanSurfer 3) Configure CRM software Install and configure Microsoft SQL2000 DB, support packs, kernel updates, Java, create SAP Central System, add Dialog Instances as needed 4) Configure benchmarking software Install CRM benchmark client and server software Diagram 1 shows the hardware architecture of the CRM Project. The diagram depicts a two-tier architecture where the database and applications run on one central host, and the SAP front-end runs on the benchmark client. The benchmark client connects to the CRM application server via a 1 Gigabit (GB) Ethernet connection. The CRM application server hands off database requests to the Microsoft SQL2000 database. The database resides on a RAID 10 LUN in the Dell/EMC storage array. The storage array and the CRM server connect via redundant QLogic 2460 Host Bus Adapters (HBA) and two Brocade 3800 Fibre channel switches. The Brocade switch is zoned so that only one initiator (HBA), and multiple targets (LUNS) exist in each zone. The Dell/EMC storage array (target device) also connects to the same physical switch and the same logical zone. EMCs PowerPath software on the CRM serevr provides optimized HBA load-balancing and failover. The storage array has multiple paths to each configured LUN. Hence, if anything was to malfunction, there are multiple safety nets to ensure that the requesting application still has access to its data. 6

Figure 1: CRM Project Architecture 7

Benchmark Findings The benchmark testing provided Dells SAP Competence Center with a good understanding of SAPs CRM environment in a2 -tier configuration, and provided several opportunities for close cooperation with SAPs technical stfaf in Walldorf, Germany. Several discoveries were analyzed to better understand the behavior of CRM and its impact on hardware. In the process, performance factors that influence response times and processor utilization in a SAP CRM environment were better understood. When performing CRM benchmarking tests, response time (saptemu) and processor utilization are measured. Several tests are run to determine which factors have the largest effect on these two measurements. Among the many tests run, the following variables provided insight into the performance of SAPs CRM software module in a 2-tier environment: Number of Instances Number of dialog work processes Sequential Memory setting Number of Instances An instance could be equivalent to an application server. However, an application server (one physical host) can have several instances running at the same time. An instance consists of a SAP run-time environment to process user requests. The run time environment includes dialog work-processes, a dispatcher, and may have update, spool, and background work-processes. Each of these processes consumes resources on the physical server and it becomes important to know how many instances should be created to maximize the server hardware and optimize the throughput. In our environment we experimented with different number of instances and recorded the response time and processor utilization. Diagram 2: Findings Number of Instances shows the discoveries that were made against a user load of 250 CRM benchmark users. 8

Diagram 2: Number of Instances Table 1: Number of Instances As the diagram shows, with 1 SAP instance in the 2-tier CRM environment, the response time was high (a slow performance of 2.734 seconds), but the CPU utilization was low (an average CPU utilization of 59%). As instances were added, the response time improved (decreased) until 3 instances, and then--at the creation of the 4th instance---- response time increased. Furthermore, the CPU utilization steadily increased as more instances were added. The sweet-spot, as far as the test environment goes, was at 3 instances providing best response time and good processor utilization. 9

Number of dialog work processes A dialog work process handles screen changes and interaction with the SAP GUI, interfacing with the end-user. By default, there are 2 dialog work-processes assigned to one instance. The maximum number of dialog work-processes is 99. Running too few dialog work processes does not leverage the available hardware. Too many may bring mission critical SAP applications to its knees. Hence it is important to find the right number of dialog work processes to optimize the environment. For the test, instances were kept at 3 and user count at 250. Dialog work-processes tested ranged from 5 to 15 Diagram 3: Number of dialog work processes With 5 dialog work-processes (DIA wp) the response time was high (a slow response time of 2.380 seconds) but the CPU utilization was low (66%). As the number of dialog work-processes increased the response time improved and processor utilization increased. For a testing scenario of 250 CRM users with 3 instances, the sweet-spot happened at 13 dialog work-processes. At 13 DIA wp the response time was a quick 0.626 seconds, and the CPU utilization was at 91 percent. Going above 13 work processes did not seem productive as both the response time and CPU utilization increased. 01

Sequential Memory setting Sequential Memory Access is enabled (in BIOS CPU settings) by default. When enabled, Sequential Memory Access optimizes the system for applications that require sequential memory access. When the setting is disabled, the memory is optimized for applications with random memory access. Sequential memory access enables the hardware to access memory passing through each intervening point. This can be compared to random memory access, which enables the hardware to access the data at random points. If data is usually accessed in the same order, sequential access may be the best setting. However, if data is accessed randomly, this setting should be disabled in BIOS. We tested both (enabled, and then disabled) in order to discover which setting provided better results for SAP CRM 5.0Below are the results. Diagram 4: Sequential Memory setting As indicated by the results, SAP CRM 5.0 performs much better if the Sequential Memory setting is turned OFF. Both response time and CPU utilization improved as this setting was disabled in BIOS. 11

Voir