R EVISION 2.0 ORACLE ENTERPRISE EXPENSES 8.8 USING DB2 FOR Z/OS 8.1 ON AN IBM z990 2084- B16 with 313 Feature [6-way LPAR] As a global leader in e-business applications, Oracle is committed to delivering high performance solutions that meet our customers expectations. Business software must deliver rich functionality with robust performance. This performance must be maintained at volumes that are representative of customer environments. Oracle benchmarks demonstrate our software s performance characteristics for a range of processing volumes in a specific configuration. Customers and prospects can use this information to determine the software, hardware, and network configurations necessary to support their processing volumes. The primary objective of our benchmarking effort is to provide as many data points as possible to support this important decision. Benchmark (English) Référence d'exécution (Français) Benchmark- Test (Deutsch) SUMMARY OF RESULTS Patrón de rendimiento (Español) Benchmark (Português) Average Response Oracle Enterprise Expenses 8.8 Extra-Large Volume Model Concurrent 2,000 Load 1.7 sec, Save 2.4 sec Oracle 8.8 Gestion des Frais Grand modèle supplémentaire de données temps de réponse Concourants Utilisateurs Antwortzeit Gleichzeitige Benutzer tiempo de reacción Simultáneos Utilizadores Oracle 8.8 Spesenabrechnung Datenbankmodell Extra-Large Load 1,7 sek, Save 2,4 sek Oracle 8.8 Gastos Volumen grande adicional de los datos tempo de resposta Despesas 8.8 do Oracle Volume grande extra dos dados Simultâneos Usuários BENCHMARK PROFILE In June 2005, Oracle (PeopleSoft) conducted a benchmark in Pleasanton, CA to measure the online performance of PeopleSoft Enterprise Expenses 8.8 using IBM DB2 for z/os 8.1 on a 6-way IBM zseries 990 model 2084-B16 with 313 Feature database server, running IBM z/os version 1.4. Two 10-way partitions and one 9-way partition of an IBM eserver pseries p690 server, running IBM AIX 5L V5.2 were used as application servers. The remaining 3-way partition was used as the web server. The 7 Terabyte (~2.8 TB available, <150 GB used) IBM Enterprise Storage Server (ESS Model 800 SHARK) was used for storage. The benchmark measured client response times for 1,700 and 2,000 concurrent users using a standard extra-large data composition model. The testing was conducted in a controlled environment with no other applications running. Tuning changes, if any, were approved by PeopleSoft Development and will be available in a future update or release. The goal of this benchmark was to obtain baseline performance data for Oracle (PeopleSoft) Expenses 8.8 Online with DB2/390 8.1 on IBM servers. The figure below illustrates average retrieve (load) and update (save) response times for a single user, and for a single user with 1,700 and 2,000 concurrent users. Average Response Time in Seconds Oracle Enterprise Expenses 8.8 Using DB2 for z/os on IBM zseries and pseries 3.0 2.5 2.0 1.5 1.0 0.5 0.0 1 User 1700 2000 Figure 1: Average Response Times Load Save * Results are weighted averages corresponding to the transaction mix specified in the Data Composition model below.
ONLINE METHODOLOGY Mercury Interactive s LoadRunner was used as the load driver, simulating concurrent users. It submitted a business process at an average rate of one every five minutes for each concurrent user. Mercury Interactive s QuickTest Professional was used to automatically submit transactions and to record the benchmark measurements on the client PC. Measurements were recorded when the user load was attained and the environment reached a steady state. Load times were measured from the time the user clicks the <OK> button until all the data for the entire business transaction has been retrieved. (Browser) 'Client' QuickTest DB Server 'Driver' LoadRunner Web Server App. SCM/ FMS 2,000 68% 3-way 73% 2 10- way, 1 9-way 37% 6-way Expense Report Weekly Entry: This scenario is included to simulate business travelers that enter all expense at once. Create an expense report to record international travel-related expenses. Navigate to the Travel and Expense Center and click on Create within the Expense Report menu. Enter 30 detail lines for international travel (i.e., per diem, auto mileage, multi-currency, etc). Include exception comments on multiple detail lines. Hotel Wizard: Navigate to the Travel and Expense Center and click on Add. Itemize a hotel charge for a 7-day business trip with five itemized hotel charges per day. Travel Authorization Weekly Entry: Create a travel authorization for international travel. Navigate to the Travel and Expense Center and click on Create within the Travel Authorization menu. Enter 4 detail lines for international travel with 4 unique expense types (i.e., airfare, per diem, varied auto mileage rates and meal dinner). Time Report Weekly Entry: Create a time report to record hours worked associated with a project. Navigate to the Travel and Expense Center and click on Create within the Time Report menu. Enter a time report with 1 unique project and 4 activities 3 hours per activity per day. Table 1 shows how the business transactions were weighted for the benchmark. The weightings are intended to simulate a typical user environment. Figure 2: Four-Tier Online Test Implementation SELF-SERVICE BUSINESS PROCESSES Oracle defines a business transaction as a series of HTML pages that guide a user through a business process, such as entering an expense report. Four business transactions within Expenses were tested for this benchmark. They are as follows: Expense Report Daily Entry: This scenario simulates those users that enter expenses daily during their business travel. Create an expense report to record domestic travelrelated expenses. Business Transaction Entry and Save Percent of Total Average Pacing (Minutes) Daily Expense Reports 25% 5 min Weekly Expense Reports with Hotel Wizard 5% 5 min Travel Authorization 10% 5 min Weekly Time Reports 60% 5 min Total 100% Table 1: Business Transaction Mix Navigate to the Travel and Expense Center and click on Create within the Expense Report menu. Enter 15 detail lines for domestic travel (i.e., airfare, hotel, and car). Include exception comments on multiple detail lines. COPYRIGHT 2005 PeopleSoft, Inc. Oracle, Inc. Page - 2
ONLINE TRANSACTION RESULTS Table 2 shows average logon, navigation, load, save and inquiry response times, in seconds, for each business transaction. Single User 1,700 2,000 Daily Expense Report Logon 1.30 1.27 2.38 SERVER PERFORMANCE Figure 3 shows the average CPU utilization for each of the servers in this test. The CPU utilization is the average across all of the CPUs in each server. Oracle Enterprise Expenses 8.8 Using DB2 for z/os on IBM zseries and pseries Nav. 0.73 0.78 0.76 Load 1.70 2.04 2.95 Save 1.81 2.42 2.83 Inquiry 1.55 1.89 1.86 Weekly Expense Report Logon 1.36 1.40 2.20 with Hotel Wizard Nav. 1.18 1.22 1.34 Load 1.73 2.08 2.65 Save 2.78 3.13 4.25 Load 2 2.60 2.86 2.99 % Average CPU Utilization 80% 70% 60% 50% 40% 30% 20% 10% 0% 1700 2000 Web Server App DB Server Travel Authorization Logon 1.31 1.69 1.40 Nav. 1.20 1.26 1.37 Load 0.81 1.05 0.99 Figure 3: Average Server CPU Utilization Save 0.64 1.14 1.68 Weekly Time Report Logon 1.33 1.41 1.72 Nav. 2.04 2.02 2.36 Load 0.60 0.70 1.03 Save 1.82 1.86 2.22 Weighted Average Load 1.13 1.35 1.69 Weighted Average Save 1.75 1.99 2.42 I/O PERFORMANCE The 2.4 Terabyte (allocated) IBM Enterprise Storage Server (ESS Model 800 SHARK) was used for storage. I/O performance is crucial to batch performance and is summarized as follows: Transactions/min n/a 340 400 Transactions/hour n/a 20,400 24,000 Avg I/Os per Sec Peak I/Os per Sec Avg. I/O Response Table 2: Online Business Transaction Runtimes 1,700 64 97 1.1 msec 2,000 104 168 1.1 msec The database and application servers were processing a total of 400 business transactions per minute at the peak load of 2,000 concurrent users. The estimated transaction rate is calculated by dividing the total number of concurrent users by the average pacing rate. Table 2: I/O Performance Performance may vary on other hardware and software platforms and with other data composition models. COPYRIGHT 2005 PeopleSoft, Inc. Oracle, Inc. All rights reserved. Page - 3
DATA COMPOSITION DESCRIPTION The database was based on our standard extra-large data model as described below: Expenses Model Size Extra-Large Model # of Business Units (US001, US006, FRA01) 25 # of SetIDs 3 # of Active Employees 25,000 # of Expense Reports processed annually 2,408,714 # of Time Reports 5,236,042 # of Travel Authorizations 3,590,831 # of Expense Report Lines 58,740,539 # of Time Report Lines 20,944,080 # of Travel Authorization Lines 14,363,299 # of Expense Report Distribution Lines 173,953,199 # of Time Report Detail Lines 104,720,420 # of Travel Authorization Detail Lines 14,363,299 # of Advances 2,394,024 # of Advance Lines 2,394,026 # of My Wallet Transactions 45,377,178 Table 3: Data Composition Twenty-four months of historical data was created for Expense Reports, Time Reports and Travel Authorizations. BENCHMARK ENVIRONMENT HARDWARE CONFIGURATION Database Server: The IBM zseries 990, model 2084-B16 with 313 Feature was used as the database server. It was equipped with the following: 6 IBM z990 Gen1 Processors (13 Processors populated, but only 6 available for this testing) 4392 MIPS total for 13 engines, 2027 MIPS for 6 engines 32 Gigabytes of Memory (3 GB available for this test) The IBM zseries 990 was attached to: One IBM Enterprise Storage Server, 2105-800 Turbo, 72.8 GB disk size, ~7 Terabytes of total Disk Space, with 2.8 Terabytes available (<150 GB used) Application Server(s): Two 10-CPU LPAR of the IBM eserver pseries p690 (7040-681) server was used as the application server. It was equipped with the following: 10 1.9 GHz POWER4 Processors, each with 32 ~41 Gigabytes of Memory A 9-CPU LPAR of the IBM eserver pseries p690 (7040-681) server was used as the application server. It was equipped with the following: 9 1.9 GHz POWER4 Processors, each with 32 ~41 Gigabytes of Memory Web Server(s): A 3-CPU LPAR of the IBM eserver pseries p690 (7040-681) server was used as the web server. It was equipped with the following: 3 1.9 GHz POWER4 Processors, each with 32 ~4 Gigabytes of Memory Load Simulation Driver: 1 IBM xseries model x235 2-way was used as a driver. It was equipped with the following: 2 2.4 GHz Intel Xeon Processors MP, each with 1 Megabyte of Level-3 Cache 4 Gigabytes of Memory COPYRIGHT 2005 PeopleSoft, Inc. Oracle, Inc. Page - 4
Client PC: SOFTWARE VERSIONS IBM Thinkpad T23 with the following: Oracle (PeopleSoft) Enterprise Expenses 8.8 1 2.0 Gigahertz Intel Xeon Processor, with 512 PeopleTools 8.45.11 Kilobytes of Level-2 Cache and 1 Megabyte of Level-3 Cache IBM DB2 for z/os 8.1 1 Gigabyte of Memory IBM z/os version 1.4 (on the Database server) IBM AIX 5L V5.2 ML02 (on the application servers and web server) Microsoft Windows XP Professional 2002 w/sp 2 (on the driver and client) Mercury Interactive's LoadRunner 7.8 Mercury Interactive s QuickTest Professional 6.5 BEA Tuxedo 8.1 RP89 with Jolt 8.1 IBM WebSphere Single Server Version 5.1 w/sp 1 with JRE 1.4.1 SR2 Merant (Micro Focus) Server Express 1.1 Oracle (PeopleSoft) Pleasanton 4500 PeopleSoft Parkway P. O. Box 8018 Pleasanton, California 94588-8618 Tel 925/694-3000 Fax 925/694-3100 Email info@peoplesoft.com World Wide Web http://www.oracle.com PeopleSoft, PeopleTools, PS/nVision, PeopleCode, PeopleBooks, PeopleTalk, and Vantive are registered trademarks, and Pure Internet Architecture, Intelligent Context Manager, and The Real-Time Enterprise are trademarks of PeopleSoft, Inc. Oracle, Inc. All other company and product names may be trademarks of their respective owners. The information contained herein is subject to change without notice. Copyright 2005 PeopleSoft, Inc. Oracle, Inc. All rights reserved. C/N 0562a-0805 IBM, the IBM logo, the eserver logo, AIX 5L, POWER4+, pseries, xseries, zseries, z/os and TotalStorage are trademarks or registered trademarks of International Business Machines, Inc. in the United States, other countries, or both. COPYRIGHT 2005 PeopleSoft, Inc. Oracle, Inc. All rights reserved. Page - 5