O RACLE R EVISION 2.6 COMPARISON OF ORACLE APPLICATION SERVER, WEBLOGIC AND WEBSPHERE USING PEOPLESOFT ENTERPRISE CAMPUS SOLUTIONS 8.9 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. SUMMARY OF RESULTS PeopleSoft Enterprise Campus Solutions 8.9 4, Concurrent Users (QuickTest Client Timers) Average Logon.96 Avg. Load Student Center 1.29 Avg. Page Load 1.7 Avg. Page Save 1.83 - Logic 1.11 1.34 1.5 1.83 - Sphere 1. 1.57 1.15 1.97 This benchmark measured the online performance for a large database model. Three web server solutions were run to investigate comparative performance issues. Testing was conducted in a controlled environment with no other applications running. The tuning changes, if any, were approved by Oracle (PeopleSoft) Enterprise Development and will be generally available in a future update. The goal of this Benchmark was to obtain comparative performance results for Oracle s, BEA s Logic and IBM s Sphere. PeopleSoft Enterprise Campus Solutions 8.9 using SQL Server on HP ProLiant and Blade Servers BENCHMARK PROFILE In February 26, Oracle (PeopleSoft) conducted a benchmark in Pleasanton, CA to measure the comparative online performance of selected processes in Oracle s PeopleSoft Enterprise Campus Solutions 8.9 w/mp 1 with Microsoft SQL Server 2 Enterprise Edition w/sp 4. We used an 8-way Hewlett-Packard ProLiant DL-74 G1 as the database server, running Microsoft Windows Server 23 Data Center Edition. Five 2-way HP ProLiant BL2p G2 Blade servers were used as Application Servers running Microsoft Windows Server 23 Enterprise Edition. A single BL2p G2 was used as the Server. An HP StorageWorks XP128 disk array was used for data Average Response Time in Seconds 3. 2. 1.. Page Load Save Logic Sphere Page Load 1.68 1.54 1.151 Save 1.825 1.825 1.965 4, Users storage. Figure 1: PeopleSoft Enterprise Campus Solutions 8.9 Response Times The weighted averages in Figure 1 correspond to the transaction proportions described in Table 3.
ONLINE METHODOLOGY Mercury Interactive LoadRunner was used as the load driver, simulating concurrent users. It submitted a business transaction at an average rate of one every 1 minutes for each concurrent user to the application servers via the web servers. Mercury Interactive 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. Figure 2 shows a typical 4-tier benchmark configuration. This benchmark was run using a physical 4 tier configuration; with the database server, the application server and the web server all being hosted on separate boxes. The Process Scheduler was also hosted on a separate server BUSINESS PROCESSES Oracle (PeopleSoft) defines a business transaction as a series of HTML pages that guide a user through a business process, such as creating a new business case. The six PeopleSoft Enterprise Campus Solutions 8.9 selfservice processes tested in this benchmark are as follows: CAMPUS SOLUTIONS Add Class: - The student logs on and navigates to the Student Center page. They click on the Add a Class link and enter the term and class id for two separate classes. The student completes the extensive enrollment options and hits Save. Calculate Tuition: - The student logs on and navigates to the Student Center page. They click on the Finances Section link and navigate to the Account Inquiry option. After viewing their balance, sign out. (Browser) 'Client' QuickTest 'Drivers' LoadRunner 4, Users Drop Class: - The student logs on and navigates to the Student Center page. They click on the Drop a Class link and enter the term and class id. The student completes the extensive enrollment options and hits Save. Server 1 2-way Load Student Center: - The student logs on and navigates to the Student Center page. They view some information then sign out. DB Server HCM App. Servers Figure 2: 4-Tier Configuration 5 2-ways 8-way Load times were measured from the time the user clicks a hyperlink or push button until the new HTML page has been rendered. Update times were measured from the time the user clicks the <SAVE> button until the new HTML page has been rendered. Measurements were recorded on all of the servers when the user load was attained and the environment reached a steady state. Make A Payment: - The student logs on and navigates to the Student Center page. They click on the Finances Section link and navigate to the Make a Payment option. Select Pay by Credit Card and enter the credit card info and payment info. Hit Save and sign out. Schedule Planner Validation: - The student logs on and navigates to the Student Center page. Navigate to the Schedule Planner page, enter the term and course numbers. Click to Validate the selected courses. Hit Save and sign out. Process % of Users Avg. Pacing Add Class 2% 1 min Calculate Tuition 15% 1 min Drop Class 1% 1 min Load Student Center 35% 1 min Make a Payment 1% 1 min Schedule Planner Validation 1% 1 min Total 1% Table 3: Business Process Mix Table 3 shows the proportions of the business processes used in the measurements of this benchmark. The proportions are intended to simulate a typical user scenario. COPYRIGHT 26 PeopleSoft, Inc. Oracle, Inc. All rights reserved. Page - 2
ONLINE PROCESS RESULTS Table 4 shows the corresponding average QuickTest response times, in onds, for each business process along with the overall averages. The percentage delta between the and Logic & Sphere runs is also shown. Process (4, Users) Add Class - WL - Logic - WS - Sphere Logon.878 8.3%.951 6.6%.936 Load SC 1.193 3.6% 1.236 46% 1.742 Page Load 1.7-5.1% 1.15 1.2% 1.83 Page Save 1.833.7% 1.846 7.% 1.961 Calculate Tuition The QuickTest results are the average of fifty iterations of each transaction on a real client. Note that the combined Page Load & Save result in the Calculate Tuition transaction is not included in the weighted average for Page Load or Page Save. The database and application servers were processing a total of ~4 business processes per minute at the peak load of 4, concurrent users. The transaction rate is calculated by dividing the number of users by the corresponding pacing. SERVER PERFORMANCE Logon.957 13.4% 1.85.1%.958 Load S C 1.3 2.1% 1.327 -.2% 1.298 Page Load & Save * Drop Class 3.52-2.4% 3.434 -.5% 3.53 Logon.926 2.1% 1.112 7.6%.996 Load SC 1.174 7.5% 1.262 7.2% 1.259 Page Load 1.156 7.1% 1.238 9.9% 1.271 Page Save 1.718 3.1% 1.771 5.8% 1.817 Load Student Center Logon 1.46 14.6% 1.199-6.%.983 Load SC 1.375 5.4% 1.449 -.1% 1.374 Make Payment Logon.841 41.1% 1.187 49.1% 1.254 Load SC 1.266 7.9% 1.366 88.8% 2.39 Page Load.838-3.2%.811 3.4% 1.93 Page Save 1.97-3.2% 1.97 9.9% 2.165 Schedule Plan Logon.949 11.2% 1.55 3.9%.986 Load SC 1.272.2% 1.275 42.9% 1.818 Page Load 1.28-1.2% 1.193 1.6% 1.227 Page Save 1.769 -.7% 1.757 8.6% 1.921 Weighted Logon.957 15.8% 1.18 4.3%.998 Weighted Load SC 1.286 4.4% 1.343 22.2% 1.571 Weighted Load Weighted Page Save Transactions / minute Page 1.68-1.3% 1.54 7.8% 1.151 1.825 % 1.825 7.7% 1.965 4 4 4 Table 4: Business Process Response Times The QuickTest results include browser rendering on a client while the LoadRunner results are the response times to the simulated client (load driver box). Avg. % CPU Utilization Avg. % CPU Utilization on Server PeopleSoft Enterprise Campus Solutions 8.9 using SQL Server on HP ProLiant and Blade Servers 8 6 4 2 Logic Sphere 49.15 41.87 52.21 App 45.2 45.43 46.7 DB 27.82 28.3 28.45 Figure 3: Average CPU Utilization PeopleSoft Enterprise Campus Solutions 8.9 using SQL Server on HP ProLiant and Blade Servers 8 6 4 2 Total CPU Logic Sphere 49.15 41.87 52.21 Java CPU 36.54 34.95 42.39 Figure 4: Server CPU Details App DB Total CPU Java CPU COPYRIGHT 26 PeopleSoft, Inc. Oracle, Inc. All rights reserved. Page - 3
Avg. Memory Utilization (MB) PeopleSoft Enterprise Campus Solutions 8.9 using SQL Server on HP ProLiant and Blade Servers 4, 3, 2, 1, Memory Logi c Sph ere 3567.54 323.34 3666.94 Figure 4: Average Memory Utilization Memory Units Accomplished 6 5 4 3 2 1 Garbage Collection Avg time GC () Sphere Logic Min time GC () Max time GC () FootPrint (GB).1989.728.96538.494446 Sphere.31231.14 5.261.499998 Logic.221.648.7853.494446 Figure 5: Garbage Collection Metrics Garbage Collection Units Accomplished 2,5 2, 1,5 1, 5 # of Minor collections # of Major collections Sphere Logic Memory collected (GB) Total time GC () 214. 1. 89.43 4.26 Sphere. 32. 76.48 99.94 Logic 1842. 8. 81.69 4.89 Figure 6: Garbage Collection Metrics Continued COPYRIGHT 26 PeopleSoft, Inc. Oracle, Inc. All rights reserved. Page - 4
DATA COMPOSITION DESCRIPTION Table 6 summarizes the structure of the target organization. TUNING Logic: Data Model Large (5) JVMs each with a 512MB heap size. Person Records 6, Enrolled Students 1, Employees 1, Student Career Records 5, Courses in Course Catalog 5, Classes in Schedule of Classes 15, Student Enrollment Records 1,, SF Accounts 2,, Item Types 2,5 ITEM_SF Lines 9,, Table 6: Selected Data Composition Note that the Large Data Model represents five year s worth of history data. Sphere: (5) JVMs each with a 512MB heap size. Thread pool set to min of 25 and max of 4. ORB Service -> Pass by reference was checked KeepAliveEnabled is set to true MaxKeepAliveConnections is set to 4 MaxKeepAliveRequests set to 45 ConnectionKeepAliveTimeout set to 3 ConnectionIOTimeout set to 6. : (5) JVMs each with a 512MB heap size. ThreadsPerChild setting to 9 OC4JCacheSize is set to 15 keepalive to 3 Additional command line options on JVM. Xms512m Xmx512m -Xss128k Disabled all CustomLog from httpd.conf Disabled access log (access-log="false") in default-website.xml Transfer log occcurences in mod_oc4j. Disabled Dynamic Monitoring Service by using command line option -Doracle.dms.sensors=none COPYRIGHT 26 PeopleSoft, Inc. Oracle, Inc. All rights reserved. Page - 5
BENCHMARK ENVIRONMENT HARDWARE CONFIGURATION A Hewlett-Packard ProLiant DL-74 G1 was used as the batch/database server. It was equipped with the following: 8 2.8 GHz Intel Xeon processors, each with 512 Kilobytes of Level 2 Cache and 1 Megabyte of Level-3 write-back cache per CPU 16 Gigabytes of Memory (~4 GB used at peak load) 1 SAN-Connected HP StorageWorks XP128 disk array with 2 fibre-channel connections ~4.6 Terabytes of total Disk Space available (64 73 GB + 2 73 GB internal disk drives), approximately 325 GB of RAID +1 storage used for this benchmark 2 Hewlett-Packard Tachyon XL2 Fibre Channel Host Bus Adapters Application Server(s): 5 HP ProLiant BL2p G2 blade servers were used as the application servers. They were equipped with the following: 2 3.6 GHz Intel Xeon processors, each with 512 Kilobytes of Level 2 Cache and 1 Megabyte of Level-3 write-back cache per CPU 3 Gigabytes of Memory ~36.4 GB of storage with an integrated SmartArray 5i Plus Controller Server(s): 1 HP ProLiant BL2p G2 blade server was used as the web server. It was equipped with the following: 2 3.6 GHz Intel Xeon processors, each with 512 Kilobytes of Level 2 Cache and 1 Megabyte of Level-3 write-back cache per CPU 3 Gigabytes of Memory ~36.4 GB of storage with an integrated SmartArray 5i Plus Controller Load Simulation Driver(s): 1 Hewlett-Packard NetServer 1p1r was used as the load driver controller. It was equipped with the following: 2 1 Gigahertz Pentium III Xeon Processors, each with 1 Megabyte of Level-2 Cache 4 Gigabytes of Memory 1 Hewlett-Packard NetServer 1p1r was used as the load driver. It was equipped with the following: 2 1 Gigahertz Pentium III Xeon Processors, each with 1 Megabyte of Level-2 Cache 4 Gigabytes of Memory SOFTWARE VERSIONS Oracle s PeopleSoft Enterprise Campus Solutions 8.9 MP1 Oracle s PeopleSoft Enterprise (PeopleTools) 8.47.1b Microsoft SQL Server 2 Enterprise Edition w/sp 4 8..239 Microsoft Windows Server 23 Enterprise Edition w/sp 1 Build 379 (64 bit) (on the database server) Microsoft Windows Server 23 Enterprise Edition (on the Application, and Process Scheduler servers) Microsoft Windows 2 Advanced Server 5. Build 2195 (on the Drivers) Microsoft Windows XP Professional w/sp 2 (on the client) Mercury Interactive LoadRunner 8. Mercury Interactive QuickTest Professional 6.5 BEA Tuxedo 8.1 RP89 with Jolt 8.1 Microsoft Internet Explorer 6. QuickTest Client PC: HP Workstation xw6 with the following: 2 2.8 GHz Intel Pentium IV Processor 248 Megabytes of Memory COPYRIGHT 26 PeopleSoft, Inc. Oracle, Inc. All rights reserved. Page - 6
Oracle Application Server () 1.1.2 ICE Tracking: Java version "1.4.2_4" Report ID: 1322265 (Resolution Id: 592864) Java(TM) 2 Runtime Environment, Standard Edition Report ID: (build 1.4.2_4-b5) Java HotSpot(TM) Client VM (build 1.4.2_4-b5, mixed Report ID: releases ) 135917 (Resolution Id: 611993) mode) BEA Logic Server 8.1 w/sp 4 Java version "1.4.2_5" Java(TM) 2 Runtime Environment, Standard Edition (build 1.4.2_5-b4) Java HotSpot(TM) Client VM (build 1.4.2_5-b4, mixed mode) IBM Sphere Application Server 5.1.1 Build a426.1 Java version "1.4.2" Java(TM) 2 Runtime Environment, Standard Edition (build 1.4.2) Classic VM (build 1.4.2, J2RE 1.4.2 IBM Windows 32 build cn142-24626 (JIT enabled: jitc)) 1329929 (Resolution Id: To be fixed in newer Oracle (PeopleSoft) Pleasanton 45 Oracle Lane P. O. Box 818 Pleasanton, California 94588-8618 Tel 925/694-3 Fax 925/694-31 Email info@peoplesoft.com WorldWidehttp://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 26 PeopleSoft, Inc. Oracle, Inc. All rights reserved. C/N 67-66 26 Hewlett-Packard, Inc. All rights reserved. HP, Hewlett-Packard, the HP logo, Integrity and HP-UX are trademarks or registered trademarks of Hewlett- Packard, Inc. in the United States and other countries. COPYRIGHT 26 PeopleSoft, Inc. Oracle, Inc. All rights reserved. Page - 7