Front cover

Draft Document for Review January 27, 2012 7:40 am REDP-4784-00

IBM Business Process Manager (BPM) 7.5 Performance Tuning and Best Practices

Learn valuable tips for tuning

Get the latest best practices

Try the example settings

IBM Business Process Management Performance Team

ibm.com/redbooks

Redpaper

Draft Document for Review January 27, 2012 7:40 am

4784edno.fm

International Technical Support Organization IBM Business Process Management (BPM) V7.5 Performance Tuning January 2012

REDP-4784-00

4784edno.fm

Draft Document for Review January 27, 2012 7:40 am

Note: Before using this information and the product it supports, read the information in Notices on page vii.

Second Edition (January 2012) This edition applies to IBM Business Process Manager 7.5, Integration Designer 7.5, and Business Monitor 7.5. This document created or updated on January 27, 2012.

Copyright International Business Machines Corporation 2012. All rights reserved. Note to U.S. Government Users Restricted Rights -- Use, duplication or disclosure restricted by GSA ADP Schedule Contract with IBM Corp.

Draft Document for Review January 27, 2012 7:40 am

4784TOC.fm

Contents
Notices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii Trademarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix The team who wrote this paper . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi Now you can become a published author, too! . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi Comments welcome. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi Stay connected to IBM Redbooks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii Chapter 1. Architecture best practices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.1 Top tuning and deployment guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.2 Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.2.1 Common Best Practices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.2.2 Process Designer Best Practices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 1.2.3 Integration Designer Best Practices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 1.3 Topology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 1.3.1 Deploy appropriate hardware . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 1.3.2 Deploy local modules in the same server . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 1.3.3 Best practices for clustering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 1.3.4 Evaluate service providers and external interfaces. . . . . . . . . . . . . . . . . . . . . . . . . 9 1.4 Business Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 1.4.1 Optimize the Topology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 1.4.2 Use a Network with Low Latency and High Bandwidth . . . . . . . . . . . . . . . . . . . . 11 1.4.3 Use a High Performing Browser . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 1.4.4 Enable Browser Caching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 1.4.5 Locate Servers Physically Near Clients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 1.4.6 Use Modern Desktop Hardware . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 1.5 Large objects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 1.5.1 Factors affecting large object size processing . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 1.5.2 Large object design patterns. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 1.6 64-bit considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 1.7 IBM Business Monitor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 1.7.1 Event processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 1.7.2 Dashboard . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 1.7.3 Database server . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Chapter 2. Development best practices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1 Shared Best Practices. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.1 Large object best practices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Process Designer Best Practices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.1 Clear variables in taskless services in a coach. . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.2 Dont use Multi Instance Loops (MILs) in the System Lane, or for batch activities 2.2.3 Use Conditional Joins only when necessary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.4 Guidelines for error handling. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.5 Use Sequential System Lane Activities Efficiently . . . . . . . . . . . . . . . . . . . . . . . . 2.2.6 Ensure the Process Center is tuned . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.7 Physically co-locate the Process Designer with the Process Center . . . . . . . . . . 2.3 Integration Developer Best Practices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.1 Business Object Parsing Mode Considerations . . . . . . . . . . . . . . . . . . . . . . . . . .
Copyright IBM Corp. 2012. All rights reserved.

17 19 19 20 20 20 21 21 22 23 23 23 23 iii

4784TOC.fm

Draft Document for Review January 27, 2012 7:40 am

2.3.2 Service Component Architecture (SCA) considerations . . . . . . . . . . . . . . . . . . . . 2.3.3 BPEL business process considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.4 Human task considerations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.5 Business process and human tasks client considerations . . . . . . . . . . . . . . . . . . 2.3.6 Transactionality considerations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.7 Invocation style considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.8 Mediation flow considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4 WebSphere InterChange Server migration considerations . . . . . . . . . . . . . . . . . . . . . . 2.5 Authoring environment considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.6 Business Space Developer Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapter 3. Performance tuning and configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1 Performance tuning methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1.1 Pick a set of reasonable initial parameter settings . . . . . . . . . . . . . . . . . . . . . . . . 3.1.2 Monitor the system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1.3 Use monitoring data to guide further tuning changes . . . . . . . . . . . . . . . . . . . . . . 3.2 Tuning checklist . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 Common tuning parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.1 Tracing and logging flags . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.2 Java tuning parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.3 MDB ActivationSpec . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.4 Thread pool sizes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.5 JMS connection pool sizes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.6 JDBC data source parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.7 Messaging engine properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.8 Run production servers in production mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4 Advanced tuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4.1 Tracing and monitoring considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4.2 Tuning for large objects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4.3 Tuning for maximum concurrency. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4.4 Messaging tuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4.5 Clustered topology tuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4.6 Web services tuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4.7 Power management tuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4.8 Set AIX threading parameters. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5 Business Process Choreographer tuning (for BPEL Business Processes). . . . . . . . . . 3.5.1 Tuning Work-Manager-based navigation for business processes . . . . . . . . . . . 3.5.2 Tuning the business process container for JMS navigation . . . . . . . . . . . . . . . . . 3.5.3 Tuning task list and process list queries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5.4 Tuning Business Process Choreographer API calls . . . . . . . . . . . . . . . . . . . . . . . 3.5.5 Tune intermediate components for concurrency . . . . . . . . . . . . . . . . . . . . . . . . . 3.6 BPMN Business Process Tuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.6.1 Database tuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.6.2 BPD Queue Size and Worker Thread Pool. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.7 WebSphere ESB tuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.7.1 Tune the database if using persistent messaging. . . . . . . . . . . . . . . . . . . . . . . . . 3.7.2 Disable event distribution for CEI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.7.3 Configure WSRR cache timeout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.8 IBM Business Monitor tuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.8.1 Configure Java heap sizes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.8.2 Configure CEI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.8.3 Configure message consumption batch size . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.8.4 Enable KPI caching. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

28 29 30 30 31 33 35 36 37 38 41 42 42 42 43 43 44 45 45 46 46 47 47 48 48 48 48 49 50 52 55 56 57 57 57 57 58 58 59 59 59 60 60 61 61 61 61 61 62 62 62 62

iv

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784TOC.fm

3.8.5 Use table-based event delivery. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.8.6 Enable the Data Movement Service . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.9 Business Space Tuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.9.1 Ensure Browser Cache is Enabled . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.9.2 Optimize Complex Task Message Performance. . . . . . . . . . . . . . . . . . . . . . . . . . 3.9.3 Tune the HTTP Server . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.9.4 Optimize Performance when not using Federation Mode . . . . . . . . . . . . . . . . . . . 3.10 Database: General tuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.10.1 Provide adequate statistics for optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.10.2 Place database log files on a fast disk subsystem . . . . . . . . . . . . . . . . . . . . . . . 3.10.3 Place logs on a separate device from the tablespace containers . . . . . . . . . . . . 3.10.4 Provide sufficient physical memory. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.10.5 Avoid double buffering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.10.6 Refine table indexes as required . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.10.7 Archive completed process instances. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.11 Database: DB2-specific tuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.11.1 Update database statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.11.2 Set buffer pool sizes correctly . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.11.3 Maintain proper table indexing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.11.4 Size log files appropriately . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.11.5 Inline LOBs to improve throughput for high volume systems . . . . . . . . . . . . . . . 3.11.6 Use SMS for tablespaces containing large objects. . . . . . . . . . . . . . . . . . . . . . . 3.11.7 Ensure that sufficient locking resources are available . . . . . . . . . . . . . . . . . . . . 3.11.8 Bound the size of the catalog cache for clustered applications . . . . . . . . . . . . . 3.11.9 (Before DB2 V9.5) Size the database heap appropriately . . . . . . . . . . . . . . . . . 3.11.10 (Before DB2 V9.7) Size the log buffer appropriately . . . . . . . . . . . . . . . . . . . . . 3.11.11 (DB2 V9.7 and later) Consider disabling current commit . . . . . . . . . . . . . . . . . 3.11.12 Recommendations for BPEL business processes in IBM BPM 7.5.0 . . . . . . . . 3.11.13 Recommendations for Business Monitor 7.5.0 . . . . . . . . . . . . . . . . . . . . . . . . . 3.12 Database: Oracle specific tuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.12.1 Update database statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.12.2 Set buffer cache sizes correctly . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.12.3 Maintain proper table indexing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.12.4 Size log files appropriately . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.12.5 Recommendations for BPEL business processes in IBM BPM 7.5.0 . . . . . . . . . 3.13 Advanced Java heap tuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.13.1 Monitoring garbage collection (GC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.13.2 Setting the heap size for most configurations . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.13.3 Setting the Heap Size when running multiple JVMs on one system . . . . . . . . . . 3.13.4 Reduce or increase heap size if OutOfMemory errors occur . . . . . . . . . . . . . . . 3.14 Tuning for WebSphere InterChange Server migrated workloads . . . . . . . . . . . . . . . . Chapter 4. Initial configuration settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1 IBM BPM Server settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1.1 Three tier configuration with Web services and a remote DB2 system (BPEL business processes) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1.2 Three tier configuration using Human Services with BPMN processes . . . . . . . . 4.1.3 Two tier configuration using file store for JMS for BPEL business processes . . . 4.2 WebSphere ESB settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.1 WebSphere ESB common settings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.2 WebSphere ESB settings for Web services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.3 WebSphere ESB settings for JMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.4 DB2 settings for JMS persistent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

62 63 63 63 68 68 69 70 70 70 70 70 71 71 71 72 72 72 73 73 74 74 75 75 76 76 76 76 77 78 78 78 79 79 79 79 80 81 82 82 83 85 86 86 89 90 91 91 92 92 92

Contents

4784TOC.fm

Draft Document for Review January 27, 2012 7:40 am

Related publications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IBM Redbooks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Online resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . How to get Redbooks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Help from IBM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

93 93 93 95 95

vi

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784spec.fm

Notices
This information was developed for products and services offered in the U.S.A. IBM may not offer the products, services, or features discussed in this document in other countries. Consult your local IBM representative for information on the products and services currently available in your area. Any reference to an IBM product, program, or service is not intended to state or imply that only that IBM product, program, or service may be used. Any functionally equivalent product, program, or service that does not infringe any IBM intellectual property right may be used instead. However, it is the user's responsibility to evaluate and verify the operation of any non-IBM product, program, or service. IBM may have patents or pending patent applications covering subject matter described in this document. The furnishing of this document does not give you any license to these patents. You can send license inquiries, in writing, to: IBM Director of Licensing, IBM Corporation, North Castle Drive, Armonk, NY 10504-1785 U.S.A. The following paragraph does not apply to the United Kingdom or any other country where such provisions are inconsistent with local law: INTERNATIONAL BUSINESS MACHINES CORPORATION PROVIDES THIS PUBLICATION "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESS OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF NON-INFRINGEMENT, MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Some states do not allow disclaimer of express or implied warranties in certain transactions, therefore, this statement may not apply to you. This information could include technical inaccuracies or typographical errors. Changes are periodically made to the information herein; these changes will be incorporated in new editions of the publication. IBM may make improvements and/or changes in the product(s) and/or the program(s) described in this publication at any time without notice. Any references in this information to non-IBM Web sites are provided for convenience only and do not in any manner serve as an endorsement of those Web sites. The materials at those Web sites are not part of the materials for this IBM product and use of those Web sites is at your own risk. IBM may use or distribute any of the information you supply in any way it believes appropriate without incurring any obligation to you. Information concerning non-IBM products was obtained from the suppliers of those products, their published announcements or other publicly available sources. IBM has not tested those products and cannot confirm the accuracy of performance, compatibility or any other claims related to non-IBM products. Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products. This information contains examples of data and reports used in daily business operations. To illustrate them as completely as possible, the examples include the names of individuals, companies, brands, and products. All of these names are fictitious and any similarity to the names and addresses used by an actual business enterprise is entirely coincidental. COPYRIGHT LICENSE: This information contains sample application programs in source language, which illustrate programming techniques on various operating platforms. You may copy, modify, and distribute these sample programs in any form without payment to IBM, for the purposes of developing, using, marketing or distributing application programs conforming to the application programming interface for the operating platform for which the sample programs are written. These examples have not been thoroughly tested under all conditions. IBM, therefore, cannot guarantee or imply reliability, serviceability, or function of these programs.

Copyright IBM Corp. 2012. All rights reserved.

vii

4784spec.fm

Draft Document for Review January 27, 2012 7:40 am

Trademarks
IBM, the IBM logo, and ibm.com are trademarks or registered trademarks of International Business Machines Corporation in the United States, other countries, or both. These and other IBM trademarked terms are marked on their first occurrence in this information with the appropriate symbol ( or ), indicating US registered or common law trademarks owned by IBM at the time this information was published. Such trademarks may also be registered or common law trademarks in other countries. A current list of IBM trademarks is available on the Web at http://www.ibm.com/legal/copytrade.shtml The following terms are trademarks of the International Business Machines Corporation in the United States, other countries, or both:
AIX alphaWorks CICS Cognos DB2 developerWorks IBM IMS MQSeries POWER6 Redbooks Redpaper Redbooks (logo) Tivoli WebSphere z/OS

The following terms are trademarks of other companies: Intel, Intel logo, Intel Inside logo, and Intel Centrino logo are trademarks or registered trademarks of Intel Corporation or its subsidiaries in the United States and other countries. Windows, and the Windows logo are trademarks of Microsoft Corporation in the United States, other countries, or both. Java, and all Java-based trademarks and logos are trademarks or registered trademarks of Oracle and/or its affiliates. UNIX is a registered trademark of The Open Group in the United States and other countries. Linux is a trademark of Linus Torvalds in the United States, other countries, or both. Other company, product, or service names may be trademarks or service marks of others.

viii

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784pref.fm

Preface
This IBM Redpaper publication was produced by the IBM Business Process Manager (BPM) performance team. It provides performance tuning tips and best practices for IBM BPM 7.5 (all editions) and Business Monitor 7.5. These products represent an integrated development and runtime environment based on a key set of service-oriented architecture (SOA) and business process management (BPM) technologies: Service Component Architecture (SCA), Service Data Object (SDO), Business Process Execution Language for Web Services (BPEL), and Business Processing Modeling Notation (BPMN). This paper is aimed at a wide variety of groups, both within IBM (development, services, technical sales, and so forth) and customers. For those who are either considering or are in the early stages of implementing a solution incorporating these products, this document should prove a useful reference, both in terms of best practices during application development and deployment, and as a reference for setup, tuning, and configuration information. This paper provides a useful introduction to many of the issues influencing each product's performance, and can serve as a guide for making rational first choices in terms of configuration and performance settings. Similarly, those who have already implemented a solution using these products might use the information presented here to gain insight as to how their overall integrated solution performance might be improved. All of these products build on the core capabilities of the WebSphere Application Server infrastructure, so BPM solutions also benefit from tuning, configuration, and best practices information for WebSphere Application Server and the corresponding platform Java Virtual Machines (JVMs). Pointers to this information can be found in Related publications on page 89. The reader is encouraged to use this paper in conjunction with these references.

Products covered by this publication

The following list describes each product covered in this paper: IBM Business Process Manager 7.5.0 (All Editions) IBM Business Process Manager combines simplicity, ease-of-use and task management capabilities with support for enterprise integration and transaction process management requirements as part of an overall Services Oriented Architecture (SOA). Optimizes business processes by providing visibility to all process participants, fostering greater collaboration and enabling continuous process improvement across distributed platforms as well as z/OS. Increases efficiency with a federated view for performing tasks, managing work items, tracking performance and responding to eventsall in real time. Empowers knowledge workers with real time analytics to optimize business processes. Enhances time-to-value through business user focused design capabilities, including process coaching to guide users easily through the steps of a process. Confidently manage change with a unified model-driven environment that provides everybody visibility of the same process version Combines simplicity, ease-of-use and task management capabilities with from WebSphere Lombardi Edition with key capabilities from WebSphere Process Server (Advanced Edition only) to support enterprise integration and transaction process management requirements as part of an overall Services Oriented Architecture (SOA).
Copyright IBM Corp. 2012. All rights reserved.

ix

4784pref.fm

Draft Document for Review January 27, 2012 7:40 am

Backward-compatible with the latest versions of WebSphere Process Server (Advanced Edition only) and WebSphere Lombardi Edition. IBM Business Monitor 7.5 IBM Business Monitor provides comprehensive business activity monitoring (BAM), enabling users visibility into real-time, end-to-end business operations, transactions, and processes to help optimize processes and increase efficiency. Provides a high-performance business activity monitoring solution for processes and applications running in disparate environments which may or may not be implemented using any BPM technology. Built-in tools and runtime support for integrated Business Activity Monitoring of IBM Business Process Manager Fully integrated Cognos Business Intelligence Server 10.1.1 for advanced analysis and reporting on historical data Automatic generation of dashboards for your Business Process Modeling Notation 2.0 (BPMN2) processes, with real-time visibility to process instances, key performance indicators (KPIs), Cognos reports, and annotated BPMN2 process diagrams Fine-grained security to enable or prevent anyone to see a wide range of information depth or detail Enhanced business user customization of data filtering and dashboard controls & reports. Enable views of KPIs, metrics, and alerts through Web interfaces, iPad, mobile devices, and corporate portals. Available for distributed platform and z/OS.

Publication structure
The first three chapters of this publication are about best practices and tuning considerations for three different phases of IBM BPM projects: Architecture Development Deployment At least one of these chapters will be of interest to any reader of this publication, and many will find value in all three chapters. There is a list of key tuning and deployment guidelines in 1.1, Top tuning and deployment guidelines on page 2. We urge all readers to take note of this list because the authors have seen numerous instances where this information is useful. Chapter 4, Initial configuration settings on page 81 describes configuration options for representative performance workloads. The publication concludes with a list of useful references in Related publications on page 89. Below is a summary of each chapter: Chapter 1, Architecture best practices on page 1 This chapter provides recommendations for architecture and topology decisions that produce well-performing and scalable solutions. Chapter 2, Development best practices on page 17 This chapter presents guidelines for solution developers that will lead to high-performing systems. Chapter 3, Performance tuning and configuration on page 39

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784pref.fm

This chapter presents a discussion of the configuration parameters and settings for the major software components that comprise a business process management solution. Chapter 4, Initial configuration settings on page 81 This chapter provides details of the software configurations used for representative workloads used by the IBM performance team working on these products. Related publications on page 89 This chapter links to best practices, performance information, and product information for both the products discussed in this publication, and related products such as WebSphere Application Server, DB2, and so on.

The team who wrote this paper

This paper is authored by the IBM BPM performance team, with members in Austin Texas, Bblingen Germany, and Hursley England. Mike Collins Weiming Gu Paul Harris Ben Hoflich Kean Kuiper Sam Massey Dominik Sebastian Meyer Rachel Norris Chris Richardson Randall Theobald

Now you can become a published author, too!

Here's an opportunity to spotlight your skills, grow your career, and become a published author - all at the same time! Join an ITSO residency project and help write a book in your area of expertise, while honing your experience using leading-edge technologies. Your efforts will help to increase product acceptance and customer satisfaction, as you expand your network of technical contacts and relationships. Residencies run from two to six weeks in length, and you can participate either in person or as a remote resident working from your home base. Find out more about the residency program, browse the residency index, and apply online at: ibm.com/redbooks/residencies.html

Comments welcome
Your comments are important to us! We want our papers to be as helpful as possible. Send us your comments about this paper or other IBM Redbooks publications in one of the following ways: Use the online Contact us review Redbooks form found at: ibm.com/redbooks

Preface

xi

4784pref.fm

Draft Document for Review January 27, 2012 7:40 am

Send your comments in an e-mail to: [email protected] Mail your comments to: IBM Corporation, International Technical Support Organization Dept. HYTD Mail Station P099 2455 South Road Poughkeepsie, NY 12601-5400

Stay connected to IBM Redbooks

Find us on Facebook: http://www.facebook.com/pages/IBM-Redbooks/178023492563?ref=ts Follow us on twitter: http://twitter.com/ibmredbooks Look for us on LinkedIn: http://www.linkedin.com/groups?home=&gid=2130806 Explore new Redbooks publications, residencies, and workshops with the IBM Redbooks weekly newsletter: https://www.redbooks.ibm.com/Redbooks.nsf/subscribe?OpenForm Stay current on recent Redbooks publications with RSS Feeds: http://www.redbooks.ibm.com/rss.html

xii

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch01.fm

Chapter 1.

Architecture best practices

This chapter provides guidance on how to architect a high-performing and scalable IBM BPM solution. The purpose of this chapter is to highlight the best practices associated specifically with the technologies and features delivered in the IBM BPM products covered in this IBM Redpaper publication. However, these products are built on top of existing technologies (such as WebSphere Application Server and DB2). Each of these technologies has associated best practices that apply. It is not our intent to enumerate these. The reader is referred to Related publications on page 89 for a set of references and pointers to this information.

Copyright IBM Corp. 2012. All rights reserved.

4784ch01.fm

Draft Document for Review January 27, 2012 7:40 am

1.1 Top tuning and deployment guidelines

The remainder of this chapter details architectural best practices for IBM BPM 7.5.0 solutions. Development best practices and performance tuning and configuration are covered in subsequent chapters. The reader is strongly encouraged to read these chapters, because the authors have found this information to be beneficial for numerous customers over the years. If you read nothing else in this document, read and adhere to the following key tuning and deployment guidelines, because they are relevant in virtually all performance sensitive customer engagements: Use a high performance disk subsystem. In virtually any realistic topology, a server-class disk subsystem (for example, a RAID adapter with multiple physical disks) is required on the tiers that host the message and data stores to achieve acceptable performance. This point cannot be overstated. The authors have seen many cases where the overall performance of a solution is improved by several factors by utilizing appropriate disk subsystems. Utilize a 64-bit JVM, and set an appropriate Java heap size to deliver optimal throughput and response time. JVM verbosegc output helps in determining the optimal settings. Further information is available in 3.3.2, Java tuning parameters on page 43. Use a production quality database such as DB2. DB2 is a high-performing, industrial strength database designed to handle high levels of throughput and concurrency. It scales well and delivers excellent response time. Tune your database for optimal performance. Proper tuning and deployment choices for databases can greatly increase overall system throughput. For details, see 3.10, Database: General tuning on page 67. Disable tracing. Tracing is important when debugging, but the overhead of tracing severely impacts performance. More information is available in 3.4.1, Tracing and monitoring considerations on page 46. Configure thread and connection pools to enable sufficient concurrency. This is important for high volume, highly concurrent workloads, because the thread pool settings directly influence how much work can be concurrently processed by the server. For more information, see Configure thread pool sizes on page 49. Physically co-locate the Process Designer clients with the Process Center server(s). The Process Designer communicates very frequently with the Process Center, so minimizing network latency is essential. For BPEL business processes: Where possible, utilize non-interruptible processes (microflows) instead of long running processes (macroflows). Macroflows are required for many processes (for example, if human tasks are employed, or state needs to be persisted). However, there is significant performance overhead associated with macroflows. If macroflows are needed for some portion of the solution, separate the solution into both microflows and macroflows to maximize utilization of microflows. For details, see Choose non-interruptible processes whenever possible on page 5. For task and process list queries, use composite query tables. Query tables are designed to produce excellent response times for high-volume task and process list queries. For details, see Choose query tables for task list and process list queries on page 5. Use work-manager-based navigation to improve throughput for long running processes. This optimization reduces the number of objects allocated, the number of objects retrieved from the database, and the number of messages sent for Business 2
IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch01.fm

Process Choreographer messaging. For further information, see 3.5.1, Tuning Work-Manager-based navigation for business processes on page 55. Avoid unnecessary use of asynchronous invocations. Asynchronous invocation is often needed on the edges of modules, but not within a module. Utilize synchronous preferred interaction styles, as described in Set the preferred interaction style to Synchronous whenever possible on page 32. Avoid overly granular transaction boundaries in SCA and BPEL. Every transaction commit results in expensive database and messaging operations. Design your transactions with care, as described in 2.3.6, Transactionality considerations on page 30.

1.2 Modeling
This section describes best practices for modeling. IBM Business Process Manager (Advanced Edition) offers two authoring environments. IBM Process Designer (hereafter called Process Designer) is used to model and execute high-level BPMN business processes, which often have human interactions. Note that this is the only authoring tool for IBM BPM 7.5.0 Standard Edition. IBM Integration Designer (hereafter called Integration Designer) is leveraged to build and implement services that are completely automated or that invoke other services such as web services, enterprise resource applications, or applications running in CICS and IMS, which already exist in the enterprise. It is also the tool to use to author BPEL business processes. Note that these authoring environments both interact with the Process Center, which is a shared repository and runtime environment. There are two roles and skill sets to consider when developing business process management (BPM) applications using these environments: The business author is responsible for authoring all business processes. The business author is able to use services, but is not interested in the implementation details or how they work. The business author uses Process Designer to create business process diagrams (BPDs), and advanced integration services (AISs) to collaborate with the integration programmer. The integration programmer is responsible for doing all of the integration work necessary to support the processes the business author creates. For example, the integration programmer will implement all the AISs, and will produce mappings between backend formats and the requirements of current applications. The integration programmer uses Integration Designer. The rest of this section is organized based on the type of user, with separate sections describing common best practices, Process Designer (business authors) best practices, and Integration Designer (integration programmer) best practices.

1.2.1 Common Best Practices

Choose the appropriate granularity for a process
A business process and its individual steps should have business significance and not try to mimic programming-level granularity. Use programming techniques such as Plain Old Java Objects (POJOs) or Java snippets for logic without business significance. This topic is

Chapter 1. Architecture best practices

4784ch01.fm

Draft Document for Review January 27, 2012 7:40 am

discussed further in the developerWorks article Software components: coarse-grained versus fine-grained, available at the following website: http://www.ibm.com/developerworks/webservices/library/ws-soa-granularity/

Use events judiciously

The purpose of Common Base Event (CBE) emission in IBM BPM 7.5.0 is for business activity monitoring. Because CBE emission uses a persistent mechanism, it is inherently heavyweight. One should utilize CBE for events that have business relevance only. Emitting CBEs to a database is not recommended. Instead, CBE emission should be done through the messaging infrastructure. Do not confuse business activity monitoring and IT monitoring, The Performance Monitoring Infrastructure (PMI) is far more appropriate for the latter. With this in mind, the following concepts generally hold for most customers: Customers are concerned about the state of their business and their processes. Therefore, events that signify changes in state are important. For long-running and human task activities, this is fairly natural: use events to track when long-running activities complete, when human tasks change state, and so forth. For short running flows that complete within seconds, it is usually sufficient to know that a flow completed, perhaps with the associated data. It usually makes no sense to distinguish events within a microflow that are only milliseconds or seconds apart. Therefore, two events (start, end) are usually sufficient for a microflow.

1.2.2 Process Designer Best Practices

Physically co-locate the Process Designer with the Process Center
The Process Designer interacts very frequently with the Process Center for authoring tasks. As such, network latency should be minimized to provide optimal response times. Locate the Process Center in the same physical location as the Process Designer users. If this is not possible, the Process Designer users should remotely connect to the environment where the Process Center is physically located, and use the Process Designer via that mechanism.

Develop efficient tasks

Here are guidelines for developing efficient tasks: Where possible, call system lane tasks with Service No Task so the tasks are not written to the database. Limit business data search to only those fields you know you need to be searchable in the process portal.

Manage variable usage

Variables are persisted to the database when execution contexts are saved, which happens fairly frequently (that is, when transitioning from BPD to service execution and when executing each coach). These persistence operations are expensive. Minimize the persistence cost by: Minimizing the number of variables used Minimizing the size of each variable Setting variables to null when no longer needed (DB result sets are a good example of this)

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch01.fm

Turn off Auto Tracking in BPDs if it is not required

Auto Tracking is enabled by default for BPDs. This capability is very important for many BPDs since it enables the gathering, tracking, and reporting of key business metrics. However, it is an expensive operation since the events are processed by the Performance Data Warehouse and persisted in the database. Disable Auto Tracking for BPDs that do not require tracking and reporting business metrics.

Use long running BPDs sparingly

Some BPDs run forever. While this is required for certain business processes, only design these type of BPDs if the capability is strictly required. BPDs that run forever will continually poll for new events, which will utilize server CPU. Consider using an alternative communication mechanisms (such as JMS queues) instead of polling. Also, disable Auto Tracking for these BPDs to avoid excessive traffic to the Performance Data Warehouse.

Develop efficient Coaches

Here are some guidelines for developing well-performing Coaches: Use custom visibility sparingly Avoid large, complex coaches Avoid large, repeating tables, page the results instead

Minimize use of large Javascript scripts

Only a single instance of a given Javascript script can run at once (this is done to ensure functional correctness). If concurrent execution of a large script is required, break the script up into smaller pieces to obtain better parallel execution.

1.2.3 Integration Designer Best Practices

Choose non-interruptible processes whenever possible
Use interruptible processes (also known as macroflows or long running processes) only when required (for example, long running service invocations and human tasks). Non-interruptible processes (also known as microflows or short running processes) exhibit much better performance at runtime. A non-interruptible process instance is executed in one J2EE transaction with no persistence of state, while an interruptible process instance is typically executed in several J2EE transactions, requiring that state be persisted in a database at transaction boundaries. Whenever possible, utilize synchronous interactions for non-interruptible processes. A non-interruptible process is more efficient than an interruptible process because it does not have to utilize state or persistence in the backing database system. A process is interruptible if the Process is long-running checkbox is selected in the IBM Integration Developer. Navigate to Properties Details to make this selection. If interruptible processes are required for some capabilities, separate the processes so that the most frequent scenarios can be executed in non-interruptible processes and exceptional cases are handled in interruptible processes.

Choose query tables for task list and process list queries
Query tables are designed to provide good response times for high-volume task lists and process list queries. Query tables offer improved query performance: Improved access to work items reduces the complexity of the database query.
Chapter 1. Architecture best practices

4784ch01.fm

Draft Document for Review January 27, 2012 7:40 am

Configurable high-performance filters on tasks, process instances, and work items allow for efficient filtering. Composite query tables can be configured to bypass authorization through work items. Composite query tables allow the definition of query tables that reflect the information that is displayed on task lists and process lists presented to users. For further information, see the references below: WebSphere Process Server Query Table Builder http://www.ibm.com/support/docview.wss?uid=swg24021440 Query Tables in Business Process Choreography in the IBM BPM 7.5 Information Center: http://publib.boulder.ibm.com/infocenter/dmndhelp/v7r5mx/topic/com.ibm.wbpm.mai n.doc/ic-homepage-bpm.html

Choose efficient metadata management

This section describes best practices for metadata usage.

Follow Java language specification for complex data type names

While WebSphere Process Server allows characters in business object type names that would not be permissible in Java class names (the underscore, _, for example), the internal data representation of complex data type names makes use of Java types. As such, performance is better if business object types follow the Java naming standards, because if valid Java naming syntax is used, no additional translation is required.

Avoid use of anonymous derived types in XSDs

Some XSD features (restrictions on the primitive string type, for example) result in modifications to the type that require a new sub-type to be generated. If these types are not explicitly declared, a new sub-type (a derived type) is generated at runtime. Performance is generally better if this can be avoided. So, avoid adding restrictions to elements of primitive type where possible. If a restriction is unavoidable, consider creating a new, concrete SimpleType that extends the primitive type to include the restriction. Then XSD elements might utilize that type without degraded performance.

Avoid referencing elements from one XSD in another XSD

For example, if A.xsd defines an element, AElement (shown in Example 1-1), it might be referenced from another file, B.xsd (shown in Example 1-2).
Example 1-1 AElement XSD

<xs:element name="AElement"> <xs:simpleType name="AElementType"> <xs:restriction base="xs:string"> <xs:minLength value="0" /> <xs:maxLength value="8" /> </xs:restriction> </xs:simpleType> </xs:element>
Example 1-2 AElement referenced from another file

<xs:element ref="AElement" minOccurs="0" />

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch01.fm

This has been shown to perform poorly. It is better to define the type concretely and make any new elements use this type. So, A.xsd becomes what is shown in Example 1-3.
Example 1-3 AElementType XSD

<xs:simpleType name="AElementType"> <xs:restriction base="xs:string"> <xs:minLength value="0" /> <xs:maxLength value="8" /> </xs:restriction> </xs:simpleType> B.xsd becomes what is shown in Example 1-4.
Example 1-4 BElement XSD

<xs:element name="BElement" type="AElementType" minOccurs="0" />

Reuse data object type metadata where possible

Within application code, it is common to refer to types; for instance, when creating a new business object. It is possible to refer to a business object type by name (for instance in the method DataFactory.create(String uri, String typeName)). It is also possible to refer to the type by a direct reference, as in the method DataFactory.create(Type type). In cases where a Type is likely to be used more than once, it is usually faster to retain the Type (for instance, through DataObject.getType()) and reuse that type for the second and future uses.

Considerations when choosing between business state machines

Business state machines (BSMs) provide an attractive way of implementing business flow logic. For some applications, it is more intuitive to model the business logic as a state machine, making the resultant artifacts are easy to understand. However, a BSM is implemented using the business process infrastructure, so there is always a performance impact when choosing BSM over business processes. If an application can be modeled using either BSM or business processes and performance is a differentiating factor, choose business processes. There are also more options available for optimizing business process performance than there are for BSM performance.

Minimize state transitions in BSMs

Where possible, minimize external events to drive state transitions in BSMs. External event-driven state transitions are costly from a performance perspective. In fact, the total time taken to execute a BSM is proportional to the number of state transitions that occur during the life span of the BSM. For example, if a state machine transitions through states A B B B C, (four transitions), it is twice as time consuming as making transitions through states A B C (two transitions). Take this into consideration when designing a BSM. Also, automatic state transitions are much less costly than event driven state transitions.

1.3 Topology
In this section we discuss choosing an appropriate topology for your solution.

Chapter 1. Architecture best practices

4784ch01.fm

Draft Document for Review January 27, 2012 7:40 am

1.3.1 Deploy appropriate hardware

It is important to pick a hardware configuration that contains the resources necessary to achieve high performance in an IBM BPM environment. Here are some key considerations in picking a hardware configuration: Cores Ensure that IBM BPM servers (Process Servers and Process Centers) are installed on a modern server system with multiple cores. Both products scale well, both vertically in terms of symmetric multiprocessing (SMP) scaling, and horizontally, in terms of clustering. Memory IBM BPM Servers benefit from both a robust memory subsystem as well as an ample amount of physical memory. Ensure that the chosen system has server-class memory controllers and as large as possible L2 and L3 caches (optimally, use a system with at least a 4 MB L3 cache). Make sure there is enough physical memory for all the applications (JVMs) combined that are expected to run concurrently on the system. For 64-bit JVMs (the recommended configuration), 4 GB per JVM is needed if the maximum heap size is 3 GB or less. Add additional physical memory for heap sizes larger than 3 GB. When using a 32 bit JVM, a rough guideline is 2 GB per JVM instance. Disk Ensure that the systems hosting the message and data stores (typically, the database tiers) have fast storage. This means utilizing RAID adapters with writeback caches and disk arrays with many physical drives. Network Ensure that the network is sufficiently fast enough not to be a system bottleneck. As an example, a dedicated Gigabit Ethernet network is a good choice. Virtualization Take care when using virtualization such as AIX dynamic logical partitioning or VMWare virtual machines. Ensure sufficient processor, memory, and I/O resources are allocated to each virtual machine or LPAR. Avoid over-committing resources.

1.3.2 Deploy local modules in the same server

If planning to deploy modules on the same physical server, better performance is achieved by deploying the modules to the same application server JVM, as this allows the server to exploit this locality.

1.3.3 Best practices for clustering

We highly recommend the IBM Redbooks publication IBM Business Process Manager V7.5.0 Production Topologies to our readers. It is a comprehensive guide to selecting appropriate topologies for both scalability and high-availability. Here is the link: http://www.redbooks.ibm.com/redbooks/pdfs/sg247976.pdf It is not the intent of this section to repeat content from that book. Rather, we distill some of the key considerations when trying to scale up a topology for maximum performance.

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch01.fm

Use the remote messaging and remote support deployment pattern

Use the remote messaging and remote support deployment environment pattern for maximum flexibility in scaling. For more information, see the section titled Topologies and Deployment environment patterns in the IBM Business Process Manager Advanced Installation Guide at the following website: ftp://ftp.software.ibm.com/software/integration/business-process-manager/library/i muc_ebpm_dist_pdf.pdf This topology (formerly known as the Gold Topology) prescribes the use of separate clusters for applications, messaging engines, and support applications (such as the CEI (Common Event Infrastructure) server, and the Business Rules Manager). This allows independent control of resources to support the load on each of these elements of the infrastructure. Note: As with many system choices, flexibility comes with some cost. For example, synchronous CBE emission between an application and the CEI server in this topology is a remote call, which is heavier than a local call. The benefit is the independent ability to scale the application and support cluster. We assume the reader is familiar with these kinds of system tradeoffs, as they occur in most server middleware.

Single server versus clustered topology considerations

In general, there are two primary issues to consider when evaluating whether to move to a clustered topology from a single server configuration: Scalability and load balancing to improve overall performance and throughput High availability by using failover to another cluster member to prevent loss of service due to hardware or software failures. Although not mutually exclusive, there are considerations applicable to each. In this paper, the focus is on the performance (throughput) aspects of clustering, and not on the high availability aspects. Most single server workloads that are driving resources to saturation can benefit by moving to a clustered topology.

1.3.4 Evaluate service providers and external interfaces

One of the typical usage patterns for IBM BPM 7.5.0 is as an integration layer between incoming requests and backend systems for the business (target applications or service providers). In these scenarios, the throughput is limited by the layer with the lowest throughput capacity. Consider the simple case where there is only one target application. The IBM BPM-based integration solution cannot achieve throughput rates higher than the throughput capacity of the target application regardless of the efficiency of the IBM BPM 7.5.0-based implementation or the size or speed of the system hosting it. Thus, it is critical to understand the throughput capacity of all target applications and service providers, and apply this information when designing the end-to-end solution. There are two key aspects of the throughput capacity of a target application or service provider: Response time, both for typical cases and exception cases Number of requests that the target application can process at the same time (concurrency) If each of these performance aspects of the target applications can be established, a rough estimate of the maximum throughput capacity can be calculated. Similarly, if average throughput is known, either one of these two aspects can be roughly calculated as well. For
Chapter 1. Architecture best practices

4784ch01.fm

Draft Document for Review January 27, 2012 7:40 am

example, a target application that can process 10 requests per second with an average response time of 1 second can process approximately 10 requests at the same time (throughput / response time = concurrency). The throughput capacity of target applications is critical to projecting the end-to-end throughput of an entire application. Also, the concurrency of target applications should be considered when tuning the concurrency levels of the upstream IBM BPM 7.5.0-based components. For example, if a target application can process 10 requests at the same time, the process server components that invoke this application should be tuned so that the simultaneous request from IBM BPM 7.5.0 at least matches the concurrency capabilities of the target. Additionally, avoid overloading target applications, because such configurations do not result in any increase in overall application throughput. For example, if 100 requests are sent to a target application that can only process 10 requests at the same time, no throughput improvement is realized versus tuning such that the number of requests made matches the concurrency capabilities of the target. For service providers that might take a long time to reply, either as part of main line processing or in exception cases, do not utilize synchronous invocations that require a response. This is to avoid tying up the business process and its resources until the service provider replies.

1.4 Business Space

This section documents best practices for architecting, developing, and deploying Business Space (BSpace) Solutions.

1.4.1 Optimize the Topology

The performance of AJAX applications like BSpace can be broken down to a 4-tier model, shown below. Each of these tiers needs to be optimized to deliver a high performing solution. Several details for optimizing the topology are discussed further below in this paper, but for context an initial discussion is presented here.

Application Cluster Dmgr Server Messaging Cluster Support Cluster

Browser

Network

HTTP Server

DB Server

Backend/ Databases

Servers

Figure 1-1 4 tiers of AJAX performance

1. Browser AJAX applications are, by definition, applications that perform work (to some extent) on the client-side, inside the browser. All typical client work, like building the UI, is done on the client side, which differentiates these applications from classical web applications, where only the rendering of HTML is done in the browser. As discussed elsewhere in this 10
IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch01.fm

paper, optimizing the browser and client system are key to delivering excellent performance. 2. Network In contrast to static web pages, AJAX applications load the content of a page dynamically as needed instead of loading the complete page at once, which means that instead of a few (or one) requests with a big payload, AJAX applications often send many requests with smaller payloads. Hence, delays in the network can have a big impact on Business Space response times because they add time to each message and in the end, add up to a bigger delay for the overall page response time. Note that the illustration in Figure 1-1 on page 10 is simplified because the network also plays a role in communications between the servers and the databases and even for inter-server-communication in a clustered setup. However, due to the nature of AJAX applications, the first point to analyze for delays is generally between the browser and the servers. 3. Servers The server infrastructure is responsible for handling the requests from the clients that are attached to it, as well as for running business applications (processes, state machines, web services, and so on), and integrating back-end services. The setup of this infrastructure heavily depends on the chosen topology. The servers which play an important role for Business Space are: a. The HTTP server An HTTP server is not part of all topologies. However, for environments that aim to serve thousands of users, a HTTP server is indispensable. All static and dynamic requests from Business Space clients will arrive at the HTTP server, which can cache and consequently send the static (cached) content back to the client and route dynamic request to the WebSphere Process Server REST API. Furthermore, an HTTP server can provide load balancing and support high-availability-scenarios. b. BPM 7.5.0 Server BPM 7.5.0 Servers execute a variety of business logic (e.g. BPEL and BPMN business processes), but in the scope of this section we focus only on actions that play a role for the Business Space Widgets, such as querying task lists, creating and claiming tasks, and so on. 4. Databases Two databases often play a key role for BPM 7.5.0 BSpace Solutions: the Business Space database and the Business Process Choreographer database (BPEDB). a. Business Space database The Business Space database holds configuration-related information and is infrequently accessed by each user. b. Business Process Choreographer database (BPEDB) All types of process- and task-information are stored in the BPEDB, which is therefore frequently updated and read when working with human task widgets.

1.4.2 Use a Network with Low Latency and High Bandwidth

As discussed, BSpace 7.5.0 is an AJAX application. As such, it utilizes multiple HTTP requests per user action, so minimizing the network delay per request is crucial. Network bandwidth is also crucial, since some HTTP requests return a significant amount of data. Where possible, utilize a high speed and high bandwidth (1 Gb or faster), dedicated network for connectivity between BSpace clients and servers.

Chapter 1. Architecture best practices

11

4784ch01.fm

Draft Document for Review January 27, 2012 7:40 am

1.4.3 Use a High Performing Browser

The choice of browser technology is absolutely crucial to BSpace performance. The choice of browsers is particularly important for Internet Explorer (IE) browsers; more recent versions of IE typically perform much better than older versions of IE.

1.4.4 Enable Browser Caching

Browsers will generally cache static data after it is initially retrieved from the server; this can significantly improve response time for scenarios once the cache is primed. This is particularly true for networks with relatively high latency. Ensure that the browser cache is active and is effective. Cache settings are browser-specific; see 3.9.1, Ensure Browser Cache is Enabled on page 61 for further details.

1.4.5 Locate Servers Physically Near Clients

One factor that influences network latency is the physical distance between servers and clients, and also between servers in a cluster (for example, between the support cluster and the BPM 7.5.0 application servers). Where practical, locate BPM 7.5.0 servers physically close to each other, and to the BSpace clients, to minimize the latency for requests.

1.4.6 Use Modern Desktop Hardware

For BSpace solutions, much of the processing is done on the client system (browser rendering), so it is imperative to deploy modern desktop hardware with high speed processors with large caches and fast front side buses, along with sufficient physical memory. Monitor your client systems with performance tools (Windows Task Manager or vmstat) to ensure that the client system has sufficient CPU and memory resources to ensure high performance.

1.5 Large objects

An issue frequently encountered by field personnel is identifying the largest object size that the IBM BPM 7.5.0 server, and corresponding WebSphere Adapters, can effectively and efficiently process. There are a number of factors affecting large object processing in each of these products. This section presents both a discussion of the issues involved and practical guidelines for the current releases of these products. Note that this issue primarily applies to 32 bit JVMs, due to the constraints on heap sizes in that environment. The single most important factor affecting large object processing is the JVM. Note that there was a significant change in JVM technology starting with Java 5, which was first used by the IBM BPM products in V6.1.0. IBM BPM 7.5.0 utilizes the Java 6 JVM, which is the follow-on release to Java 5. Due to the changes in the underlying JVM technology, this section has been rewritten and the recommendations and best practices differ from those in IBM BPM 6.0.2 and earlier, which used the 1.4.2 JVM. In general, objects 5 MB or larger might be considered large and require special attention. Objects 100 MB or larger are very large and generally require significant tuning to be processed successfully.

12

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch01.fm

1.5.1 Factors affecting large object size processing

Stated at a high level, the object size capacity for a given installation depends on the size of the Java heap and the load placed on that heap (that is, the live set) by the current level of incoming work. The larger the heap, the larger the business object that can be successfully processed. To apply this statement, one must first understand that the object size limit is based on three fundamental implementation facts of JVMs: Java heap size limitations The limit to the size of the Java heap is operating system-dependent. Further details of heap sizes are given later in Heap limitations: Increase the Java heap to its maximum on page 47, but it is not unusual to have a heap size limit of around 1.4 GB for 32-bit JVMs. The heap size limit is much higher on 64-bit JVMs, and is typically less of a gating factor on modern hardware configurations than the amount of available physical memory. Size of in-memory business objects Business objects, when represented as Java objects, are much larger in size than when represented in wire format. For example, a business object that consumes 10 MB on an input JMS message queue might result in allocations of up to 90 MB on the Java heap. The reason is that there are many allocations of large and small Java objects as the business object flows through the adapters and IBM BPM 7.5.0. There are a number of factors that affect the in-memory expansion of business objects.: The single-byte binary wire representation is generally converted to multi-byte character representations (for example, Unicode), resulting in an expansion factor of 2. The business object might contain many small elements and attributes, each requiring a few unique Java objects to represent its name, value, and other properties. Every Java object, even the smallest, has a fixed overhead due to an internal object header that is 12 bytes long on most 32 bit JVMs, and larger on 64 bit JVMs. Java objects are padded to align on 8 byte or 16 byte address boundaries. As the business object flows through the system, it might be modified or copied, and multiple copies might exist at any given time during the end-to-end transaction. This means that the Java heap must be large enough to host all these business object copies for the transaction to complete successfully. Number of concurrent objects being processed The largest object that can be successfully processed is inversely proportional to the number of requests being processed simultaneously. This is due to the fact that each request has its own memory usage profile (liveset) as it makes its way through the system. So, simultaneously processing multiple large objects dramatically increases the amount of memory required, because the sum total of each requests livesets has to fit into the configured heap.

1.5.2 Large object design patterns

The following sections describe the two proven design patterns for processing large objects successfully. In cases where neither can be applied, 64-bit mode should be considered. See 1.6, 64-bit considerations on page 14 for details. Also, for WebSphere ESB a developerWorks article is available detailing best practises and tuning for large messages at the following website:

Chapter 1. Architecture best practices

13

4784ch01.fm

Draft Document for Review January 27, 2012 7:40 am

http://www.ibm.com/developerworks/webservices/library/ws-largemessaging/

Batched inputs: Send large objects as multiple small objects

If a large object needs to be processed, the solutions engineer must find a way to limit the number of large Java objects that are allocated. The primary technique for doing this involves decomposing large business objects into smaller objects and submitting them individually. If the large objects are actually a collection of small objects, the solution is to group the smaller objects into conglomerate objects less than 1 MB in size. This has been done at several customer sites and has produced good results. If there are temporal dependencies or an all-or-nothing requirement for the individual objects, the solution becomes more complex. Implementations at customer sites have shown that dealing with this complexity is worth the effort as demonstrated by both increased performance and stability. Certain WebSphere Adapters (such as the Flat Files Adapter) can be configured to use a SplitBySize mode with a SplitCriteria set to the size of each individual object. In this case a large object would be split in chunks (of a size specified by SplitCriteria) to reduce peak memory usage.

Claim check pattern: Only a small portion of an input message is used

When the input business object is too large to be carried around in a system and there are only a few attributes that are actually needed by that process or mediation, one can exploit a pattern called the claim check pattern. The claim check pattern, as applied to a business object has the following steps: 1. Detach the data payload from the message. 2. Extract the required attributes into a smaller control business object. 3. Persist the larger data payload to a data store and store the claim check as a reference in the control business object. 4. Process the smaller control business object, which has a smaller memory footprint. 5. Where the solution needs the whole large payload again, check out the large payload from the data store using the key. 6. Delete the large payload from the data store. 7. Merge the attributes in the control business object with the large payload, taking the changed attributes in the control business object into account. The claim check pattern requires custom code and snippets in the solution. A less developer-intensive variant would be to make use of custom data bindings to generate the control business object. This approach is limited to certain export and import bindings. The full payload still must be allocated in the JVM.

1.6 64-bit considerations

IBM BPM applications can be run using either 32-bit or 64-bit JVMs. However, 64-bit mode is recommended for both Process Server and Process Center. This is bcause in 32-bit mode, the maximum heap size is limited by the 4 GB address space size. In most 32-bit operating systems, the practical limit varies between 1.5 2.5 GB. In contrast, while maximum heap size is essentially limitless in 64-bit mode, standard Java best practices still apply. The sum of the maximum heap sizes and native memory utilzation (threads, stacks, JITted code) of all the Java processes running on a system, plus additional memory required for the operating

14

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch01.fm

system and other applications, should not exceed the physical memory available on the system. IBM BPM 7.5.0 servers run most efficiently on a 64-bit JVM instance due to the much larger amount of memory that is accessible in this mode. The performance and memory footprint of a 64-bit runtime server is about the same as the 32-bit version.This was not the case for earlier versions, such as BPM v6.1 and v6.2. The following list details the factors to consider when determining in which of these modes to run: 64-bit mode is an excellent choice for applications whose liveset approaches or exceeds the 32-bit limits. Such applications either experience OutOfMemoryExceptions or suffer excessive time in GC. We consider anything greater than 10% of time in GC as excessive. These applications exhibit much better performance when allowed to run with the larger heaps they need. However, there must always be sufficient physical memory on the system to back the Java heap size. 64-bit mode is also an excellent choice for applications that, though well-behaved on 32-bit, could be algorithmically modified to perform better with larger heaps. An example would be an application that frequently persists data to a data store to avoid maintaining a very large in-memory cache, even if such a cache would greatly improve throughput. Recoding such an application to tradeoff the more space available in 64-bit heaps for less execution time yields better performance. Moving to 64-bit may cause some degradation in throughput if a 32-bit application fits well with a 1.5 - 2.5 GB heap, and the application is not expected to grow significantly. For these situations where the memory limitation is not a signifcant factor, using 32-bit JVMs may be a better choice than 64-bit.

1.7 IBM Business Monitor

This section describes best practices for performance with IBM Business Monitor 7.5.0.

1.7.1 Event processing

A major factor in event processing performance is the tuning of the Monitor database. Attention should be paid to ensure adequate bufferpool sizes to minimize disk reading activity, and the placement of the database logs, which ideally should be on a physically separate disk subsystem from the database tablespaces. By default, events are delivered from CEI to the monitor database, bypassing an intermediate queue. We recommend using this default delivery style for better performance, as it avoids an additional persistence step in the flow. For additional background see the Information Center article Bypassing the JMS Queue at the following website: http://publib.boulder.ibm.com/infocenter/dmndhelp/v7r5mx/index.jsp?topic=/com.ibm. wbpm.mon.imuc.doc/inst/cfg_qb.html

1.7.2 Dashboard
The platform requirements of the Business Space and its Monitor widgets are relatively modest compared to those of Monitor server and the database server. The most important consideration for good Dashboard performance is to size and configure the database server

Chapter 1. Architecture best practices

15

4784ch01.fm

Draft Document for Review January 27, 2012 7:40 am

correctly. Be sure it has enough CPU capacity for anticipated data mining queries, enough RAM for bufferpools and so forth, and plenty of disk arms.

1.7.3 Database server

Both event processing and Dashboard rely on a fast, well-tuned database server for good performance. The design of Monitor assumes that any customer using it has strong on-site database administrator skills. We advise that the database tuning advice and recommendations beginning in 3.10, Database: General tuning on page 67 be read and followed.

16

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch02.fm

Chapter 2.

Development best practices

This chapter discusses best practices that are relevant to the solution developer. It addresses modeling, design, and development choices that are made while designing and implementing an IBM BPM 7.5 solution. IBM Business Process Manager (Advanced Edition) offers two authoring environments. Note that these authoring environments both interact with the Process Center, which is a shared repository and runtime environment. IBM Process Designer (hereafter called Process Designer) is used to model and execute high-level business processes, which often have human interactions. Note that this is the only authoring tool for IBM BPM 7.5 Standard Edition. IBM Integration Designer (hereafter called Integration Designer) is leveraged to build and implement services that are completely automated or that invoke other services such as web services, enterprise resource applications, or applications running in CICS and IMS, which already exist in the enterprise. It is also the tool to use to author BPEL business processes. There are two roles and skill sets to consider when developing business process management (BPM) applications using these environments: The business author is responsible for authoring all business processes. The business author is able to use services, but is not interested in the implementation details or how they work. The business author uses Process Designer to create business process diagrams (BPDs), and advanced integration services (AISs) to collaborate with the integration programmer. The integration programmer is responsible for doing all of the integration work necessary to support the processes the business author creates. For example, the integration programmer will implement all the AISs, and will produce mappings between backend formats and the requirements of current applications. The integration programmer uses Integration Designer. The rest of this chapter is orgainized based on the type of user, with separate sections describing shared best practices, Process Designer (business authors) best practices, and Integration Designer (integration programmer) best practices. Additionally, developer considerations for Business Space solutions, WebSphere InterChange Server migration, and Authoring environment considerations are provided.

Copyright IBM Corp. 2012. All rights reserved.

17

4784ch02.fm

Draft Document for Review January 27, 2012 7:40 am

2.1 Shared best practices

The following Best Practices are applicable for both the Process Designer and the Integration Developer.

2.1.1 Large object best practices

This section discusses best practices for performance (particularly memory utilization) when using large objects. Very large objects put significant pressure on Java heap utilization, particularly for 32-bit JVMs. 64-bit JVMs are less succeptible to this issue due to the large heap sizes that can be configured, although througput, response time, and overall system performance can still be impacted by excessive memory usage. Follow these best practices to reduce the memory pressure and avoid OutOfMemory execeptions, and improve overall system performance.

Avoid lazy cleanup of resources

Lazy cleanup of resources adds to the liveset required when processing large objects. Any resources that can be cleaned up (for example, by dropping object references when no longer required) should be done as soon as is practical.

Avoid tracing when processing large business objects

Tracing and logging can add significant memory overhead. A typical tracing activity is to dump the business object payload. Creating a string representation of a large business object can trigger allocation of many large and small Java objects in the Java heap. Avoid turning on tracing when processing large business object payloads in production environments. Avoid constructing trace messages outside of conditional guard statement. For example, the following sample code creates a large String object even if tracing is disabled: String boTrace = bo.toString(); While this pattern is always inefficient, it hurts performance even more if the business object size is large. To avoid unnecessarily creating a business object when tracing is disabled, move the String construction inside an if statement, as is shown in the following code: if (tracing_on) System.out.println(bo.toString());

Avoid buffer-doubling code

Study the memory implications of using Java data structures that expand their capacity based on input (for example, StringBuffer and ByteArrayOutputStream). Such data structures usually double their capacity when they run out of space. This doubling can produce significant memory pressure when processing large objects. If possible, always assign an initial size to such data structures.

18

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch02.fm

2.2 Process Designer Best Practices

Following are best practices for developing high performance business processes using the Process Designer.

2.2.1 Clear variables in taskless services in a coach

Taskless services in a coach are not garbage collected until the EJB timeout occurs (two hours by default). To reduce memory utilization, set variables in the coach to null in a custom HTML block.

2.2.2 Dont use Multi Instance Loops (MILs) in the System Lane, or for batch activities
Use caution when using sub-BPDs as the activity of a multi instance loop (MIL). There is no issue if the first activity is a user task instead of a system lane. However, MILs should not be used to attempt to execute batch or system lane activities. This pattern can generate an excessive number of tokens for the BPD to process. Figure 2-1 shows an example of a good BPD design pattern.

Figure 2-1 Good BPD design pattern

Figure 2-2 shows an example of a poor BPD design pattern.

Chapter 2. Development best practices

19

4784ch02.fm

Draft Document for Review January 27, 2012 7:40 am

Figure 2-2 Poor BPD design pattern

2.2.3 Use Conditional Joins only when necessary

Simple Joins use an and condition; all lines heading into the join must have an active token for the tokens to continue forward. By contrast, for Conditional Joins all possible tokens must reach the join before they proceed. So if you have a Conditional Join with 3 incoming lines, but only 2 currently have tokens (or could have tokens by looking upstream), then those 2 tokens must arrive at the join to proceed. To determine this, the BPD engine must evaluate all possible upstream paths to determine if the tokens can arrive at the join. This can be quite expensive for large, complex BPDs. Use this capability judiciously in these cases.

2.2.4 Guidelines for error handling

Avoid global error handling in a service. This can utilize an excessive amount of server CPU, and can even result in infinite loops in Coaches. When catching errors on an activity in a BPD, dont route the error back to the same activity. This will cause the server to thrash between the BPD and service engine utilizing a large amount of CPU and database processing.

2.2.5 Use Sequential System Lane Activities Efficiently

Each system lane activity is a new Event Manager (EM) task, which adds a new task transition in the service engine. These task transitions are expensive. If your BPD contains multiple system lane service tasks in a row, use one system lane task that wraps the others to minimiize the transition overhead.

20

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch02.fm

Following is a demonstration of this issue, with Figure 2-3 showing the poor usage pattern (multiple consecutive system lane activities), and Figure 2-4 showing the more optimal usage pattern (one system lane activity that incorporates the multiple activities in the previous diagram).

Figure 2-3 Poor BPD usage pattern

Figure 2-4 Well BPD usage pattern

2.2.6 Ensure the Process Center is tuned

At a minimum, tune the following parameters for the Process Center: Use a 64-bit JVM. Use the generational concurrent garbage collector via the -Xgcpolicy:gencon JVM argument. Ensure the Java heap size is sufficiently large to handle the peak load. See 3.3.2, Java tuning parameters on page 43 for further information on setting the Java heap size. Set the ORB thread pool size to at least 20. Set the JDBC connnection pool size to at least two times the ORB thread pool size. Tune the Process Center database. See 3.10, Database: General tuning on page 67 for further information.
Chapter 2. Development best practices

21

4784ch02.fm

Draft Document for Review January 27, 2012 7:40 am

2.2.7 Physically co-locate the Process Designer with the Process Center
The Process Designer interacts very frequently with the Process Center for authoring tasks. As such, network latency should be minimized to provide optimal response times. Locate the Process Center in the same physical location as the Process Designer users. If this is not possible, the Process Designer users should remotely connect to the environment where the Process Center is physically located, and use the Process Designer via that mechanism.

2.3 Integration Developer Best Practices

Following are best practices for developing well-performing solutions via the Integration Developer.

2.3.1 Business Object Parsing Mode Considerations

This section discusses considerations for choosing the Business Object (BO) Parsing Mode. There are two options: Eager Parsing mode and Lazy Parsing mode.

Overview
WebSphere BPM version 7.001 introduced BO Lazy parsing mode. In this mode, when BOs are created their properties are not populated until they are accessed by the application. Also, when reading XML input, the stream is incrementally parsed and properties are not materialized until they are accessed. This is in contrast to Eager parsing mode, where the input XML stream is immediately fully parsed (with some exceptions for specific mediation primitives) and materialized into a complete in-memory data object representation. Lazy parsing utilizes the XML Cursor Interface (XCI) implementation of SDO provided by the WebSphere Application Server Feature Pack for SCA, which is delivered as part of IBM BPM 7.5.0. In this mode, the input XML stream is kept in memory throughout the life time of the corresponding BO. Nodes are created in memory by the cursor based XML parser only when they are traversed, and properties and attributes are evaluated only when they are requested by the application. Parsing the XML stream in this fashion delivers better performance than eager modes complete parsing for many applications. This is particularly true if only a small portion of the BO is accessed during the application execution. Lazy parsing also reuses the XML stream when BOs are serialized, e.g. for outbound service calls. This results in faster serialization since it is not necessary to serialize the entire in-memory BO into XML strings. Lazy parsing will automatically detect, and correct, if namespaces in the output stream are different than those in the original input stream. This form of serialization is used for all bindings and internal BPM runtime components, such as the Business Process Choreographer (BPC) container, and when custom code calls BO serialization. To further optimize performance, Lazy parsing also employs a technology called copy-on-write, where properties in BOs are lazily copied. When the copy operation is initially performed, the target BO just points to the source BOs node tree; no properties are copied at this point. Subsequent modifications of either the source or target BO trigger a split of the affected node in the BO tree, copying BO properties only as needed. Modifications occur in the local copy of the corresponding node. Copy-on-write is completely transparent to the application.

22

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch02.fm

Setting Parsing Mode

In BPM 7.5.0, the default parsing mode is the lazy mode for newly developed applications in IBM Integration Developer (IID) 7.5.0. Applications migrated from WPS 7.0.0.3 and earlier releases continue to execute in the Eager parsing mode. However, for these legacy applications, the Lazy parsing mode can be enabled manually via IID. IID provides a property for each module and library to configure the XML parsing mode to either eager or lazy. Figure 2-5 and Figure 2-6 on page 24 show the relevant IID panels when creating libraries and modules, respectively. When creating a library, you can specify whether it can be included in a Eager parsing or Lazy parsing module, or both. The default is set to Lazy parsing when the library is initially created. When both parsing modes are selected, the actual parsing mode at runtime will be determined by the module in which the library resides. When creating a module, you must choose either Lazy or Eager parsing mode. The default selection is the Lazy parsing mode.

Figure 2-5 Lazy parsing for new library

Chapter 2. Development best practices

23

4784ch02.fm

Draft Document for Review January 27, 2012 7:40 am

Figure 2-6 Lazy parsing for new module

In BPM 7.5.0, Lazy parsing delivers excellent performance for XML-based applications that use BOs or messages approximately 3 KB or larger, and the performance of Lazy parsing is generally better than or about equal to Eager parsing for other scenarios.

Applications benefiting from Lazy parsing

This section provides general guidance for choosing the right parsing mode, and best practices to follow when choosing one mode over the other. However, we recommend that you benchmark your application in both parsing modes to determine which mode best suits the specific characteristics of your application. The primary performance benefits derived from Lazy parsing are delayed parsing, and parsing only the properties accessed by the application. Applications exhibiting any of the following characteristics are candidates for exploiting Lazy parsing for better performance. As demonstrated in the measurements that follow, the performance improvement when using Lazy parsing can be quite significant: Only applications that use XML data streams can truly benefit from Lazy parsing. For example, applications that use Web Services bindings, synchronous or asynchronous SCA bindings, or the JMS binding with the XML data handler can potentially benefit from Lazy parsing mode. Applications with medium to large XML data streams are likely to see performance improvements when Lazy parsing is used. The performance benefit increases as the size of the XML data stream increases. In the workloads we studied, Lazy parsing outperforms Eager mode when the BOs are approximately 3 KB or larger. Clearly, the BO size will differ depending on the composition of your application, so use this as a rule of thumb. Applications that access a small portion of an XML data stream are likely to see bigger performance benefits from Lazy parsing. Access to a BO is defined as reading the value of a property, updating a property value, or adding a new property to a BO. Lazy parsing allows the unmodified portion of the data stream to pass through the system with minimal overhead. Relative to Eager parsing, the less an application accesses its BOs, the higher performance improvement it will achieve by enabling Lazy parsing.

24

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch02.fm

WESB mediations with multiple mediation primitives are likely to see performance improvements when Lazy parsing is used. When using Eager parsing, BOs are often serialized before invoking the next mediation primitive, and then deserialized once that mediation primitive is invoked. Lazy parsing mode reduces the cost of each serialization and deserialization by avoiding unnecessary materialization of unmodified properties in the data stream. While applications that fall into the sweet spots discussed above will likely benefit from Lazy parsing, often significantly, applications that dont have those attributes should see little difference in performance and can choose either Lazy or Eager parsing. For example, applications that parse non-XML data streams such as WebSphere Adapters and non-XML data handlers, applications that create BOs directly using the BOFactory service, and application that use very small XML messages (as a rule of thumb, < 3 KB) should perform similarly whether Lazy or Eager parsing mode is chosen in BPM 7.5.0. WESB mediation flows which only access the message header or which contain a single mediation primitive may see slightly better performance when using Eager parsing. For example a ServiceGateway with a single Filter mediation which routes based on the header content. However it should be noted that once multiple mediation primitives are added to a flow performance is likely to be better in Lazy parsing mode.

Minimize using mixed parsing modes between modules

Sometimes it is necessary to have some modules use Eager parsing, while others use Lazy parsing. For better performance, however, avoid using this mixed mode for application modules that frequently interact with each other. For example, if module A invokes module B through the synchronous SCA binding on each transaction, it is better to set modules A and B to both use either Eager parsing or Lazy parsing. The interaction between modules with different parsing modes is through serialization and deserialization, which can negatively affect performance. If mixed mode is unavoidable, invocation from an Eager parsing module to a Lazy parsing module is more efficient than the other way around. This is because the output from the Eager parsing module is serialized into an XML string, so the target module using Lazy parsing will still fully benefit from delayed parsing.

Use Share-by-reference libraries where possible

Often, definitions of interfaces, business objects, data maps, mediation subflows, relationships, roles, and Web service ports need to be shared so that resources in several modules can use them. The library is a project that is used to store these resources. Libraries can be deployed in several ways. They can be deployed with a process application, with the dependent module or globally. You can make your choice in the dependency editor. If you have associated a library with a process application then Process Application will be selected in the dependency editor. The library will be shared within the deployed process application, meaning only one copy will reside in memory. We call this type of library a share-by-reference library. The share-by-reference library is a new feature in BPM 7.5.0. If you choose to deploy a library with the module, which is the default setting when the library has not been associated with a process application, a copy of the library is made for each module when the module is deployed. We call this type of library a share-by-value library. More details about BPM 7.5 library types can be found inthe Libraries section in the IBM Information Center:

Chapter 2. Development best practices

25

4784ch02.fm

Draft Document for Review January 27, 2012 7:40 am

http://publib.boulder.ibm.com/infocenter/dmndhelp/v7r5mx/index.jsp?topic=%2Fcom.ib m.wbpm.wid.main.doc%2Fnewapp%2Ftopics%2Fclibrary.html Lazy Parsing is optimized to exploit share-by-reference libraries. When a BO is passed from one module to another as a parameter of a synchronous SCA service invocation, a lazy copy is used if the source and target modules share the BO schema through a share-by-reference library. Without the share-by-reference optimization, the BO is serialized on the caller side and deserialized on the callee side.

Differences in behavior for Lazy Parsing and Eager Parsing

If you are changing an application that was originally developed using Eager parsing to use Lazy parsing, or if you are planning to switch an application between Lazy and Eager parsing mode, be aware of the differences in behavior between each mode, summarized below.

Error handling
If the XML byte stream being parsed is ill-formed, parsing exceptions occur: With Eager parsing, the exceptions occur as soon as the BO is parsed from the inbound XML stream. With Lazy parsing, the parsing exceptions occur latently when the BO properties are accessed and the portion of the XML that is ill-formed is parsed. To properly handle ill-formed XML for either parsing mode, select one of the following options: Deploy a mediation flow component on the application edges to validate inbound XML. Author lazy error-detection logic at the point where BO properties are accessed.

Exception stacks and messages

Because Eager and Lazy parsing have different underlying implementations, stack traces generated by the BO programming interfaces and services have the same exception class name, but might not contain the same exception message or wrapped set of implementation-specific exception classes.

XML serialization format

Lazy parsing provides a fast serialization optimization that attempts to copy unmodified XML from the inbound data stream to the outbound data stream upon serialization. The fast serialization improves performance, but the resultant serialization format of the outbound XML data stream might be syntactically different than if the entire BO was updated in Lazy parsing, or if Eager parsing was used. Although the output XML serialization format might not be precisely syntactically equivalent across these cases, the semantics of the XML data streams are equivalent regardless of the parsing mode, and the resultant XML can be safely passed between applications running in different parsing modes with semantic equivalence.

Business object instance validator

The Lazy parsing instance validator provides a higher fidelity validation of BOs, particularly for facet validation of property values. Because of these improvements in fidelity, the Lazy parsing instance validator catches functional issues that are not caught in Eager parsing mode, and also provides more detailed error messages.

Migration of XSL stylesheets

When an application authored using WID 7.001 or earlier is migrated using IID 7.5 and rebuilt, XSL stylesheets are regenerated from the associated map files. XSL stylesheets built by IID 7.5 no longer contain whitespace or indent directives. This change was made to improve 26
IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch02.fm

performance in Lazy parsing mode. However if an XSLT primitive refers directly to a stylesheet which has been edited manually the stylesheet will not be regenerated when the application is built. In this case performance will be improved by removing the whitespace directive ( <xsl:strip-space elements="*"/> ) if it appears and by ensuring that indentation is disabled ( indent="no" ) unless specifically required.

Private APIs
The BO programming model provides support for BO APIs and a set of Service Data Object (SDO) APIs. In some cases, applications might use additional implementation-specific APIs that are not part of the supported APIs for BOs. For example, an application might use the Eclipse Modeling Framework (EMF) APIs. Although these APIs may have worked in prior BPM releases, and in Eager parsing mode, they are not supported APIs for accessing BOs. EMF APIs are considered private to the implementation of BOs and should not be used by applications. In all cases, it is strongly recommended that private APIs be removed from the application and all BO access should by done using the supported API set.

Service Message Object (SMO) EMF APIs

A mediation flow component provides the ability to manipulate message content using the Java classes and interfaces provided in the com.ibm.websphere.sibx.smobo package. For Lazy parsing, the Java interfaces in the com.ibm.websphere.sibx.smobo package can still be used, but methods that refer directly to Eclipse Modeling Framework (EMF) classes and interfaces or that are inherited from EMF interfaces are likely to fail. Also, the SMO and its contents cannot be cast to EMF objects when using Lazy parsing.

Migration
All applications developed prior to BPM 7.5 used Eager parsing mode by default. When they are migrated using the BPM runtime migration tools, they will continue to run in Eager parsing mode. To enable an application originally developed using Eager parsing to be configured to use Lazy parsing, first use the IID 7.5 to migrate the artifacts of the application. After migration, use the IID 7.5 to configure the application to use Lazy parsing. For information on migrating artifacts in IBM Integration Developer, go the following website: http://publib.boulder.ibm.com/infocenter/dmndhelp/v7r5mx/index.jsp?topic=%2Fcom.ib m.wbpm.wid.imuc.doc%2Ftopics%2Ftmigsrcartwid.html For information on setting the parsing mode, go to the following website: http://publib.boulder.ibm.com/infocenter/dmndhelp/v7r5mx/index.jsp?topic=%2Fcom.ib m.wbpm.wid.main.doc%2Fnewapp%2Ftopics%2Ftconfigbo.html

2.3.2 Service Component Architecture (SCA) considerations

This section discusses the SCA authoring considerations for performance.

Cache results of ServiceManager.locateService()

When writing Java code to locate an SCA service, either within a Java component or a Java snippet, consider caching the result for future use, as service location is a relatively expensive operation. IBM Integration Developer-generated code does not do this, so editing is required to cache the locateService result.

Chapter 2. Development best practices

27

4784ch02.fm

Draft Document for Review January 27, 2012 7:40 am

Reduce the number of SCA modules, when appropriate

IBM BPM 7.5 components are assembled into modules for deployment. When assembling modules we recognize that many factors come into play. Performance is a key factor, but maintainability, versioning requirements and module ownership must be considered as well. In addition, more modules can allow for better distribution across servers and nodes. It is important to recognize that modularization also has a cost. When components are placed together in a single server instance, it is best to package them within a single module for best performance.

Use synchronous SCA bindings across local modules

For cross-module invocations, where the modules are likely to be deployed locally (that is, within the same server JVM) we recommend using the synchronous SCA binding. This binding has been optimized for module locality and outperforms other bindings. Synchronous SCA is as expensive as other bindings when invocations are made between modules located in different IBM BPM 7.5 server instances.

Utilize multi-threaded SCA clients to achieve concurrency

Synchronous components that are invoked locally (that is, from a caller in the same server JVM) execute on the context of the callers thread. Thus concurrency, if desired, must be provided by the caller in the form of multiple threads.

Add Quality of Service qualifiers at appropriate level

Quality of Service (QoS) qualifiers such as Business Object Instance Validation can be added at the interface level, or at an operation level within an interface. Because there is additional overhead associated with QoS qualifiers, do not apply a qualifier at the interface level if it is not needed for all operations of the interface.

2.3.3 BPEL business process considerations

This section discusses considerations for authoring performance specific to BPEL business processes.

Modeling best practices for activities in a business process

Use the following guidelines when modeling a BPEL business process: Use the Audit Logging property for Business Processes only setting if you need to log events in the BPE database. This property can be set at the activity or process level. If it is set at the process level, the setting is inherited by all activities. For long-running processes, disable the Enable persistence and queries of business-relevant data property under Properties Server, for both the process and for each individual BPEL activity. Enabling this flag causes details of the execution of this activity to be stored in the BPC database. This increases the load on the database and the amount of data stored for each process instance. This setting should be used only if this specific information will need to be retrieved later. For long-running processes, a setting of participates on all activities generally provides the best throughput performance. See Avoid two-way synchronous invocation of an asynchronous target on page 31 for more details. Human tasks can be specified in business processes (for example, process administrators), invoke activities, and receive activities. Specify human tasks only if needed. When multiple users are involved, use group work items (people assignment criterion: Group) instead of individual work items for group members (people assignment criterion: Group Members). 28
IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch02.fm

Avoid 2-way synchronous invocation of long running processes

When designing long-running business process components, ensure that callers of a 2-way (request/response) interface do not use synchronous semantics, as this ties up the caller's resources (thread, transaction, and so forth) until the process completes. Instead, such processes should either be invoked asynchronously, or through a 1-way synchronous call, where no response is expected. In addition, calling a 2-way interface of a long-running business process synchronously introduces difficulties when exceptions occur. Suppose a non-interruptible process calls a long-running process using the 2-way request/response semantics, and the server fails after the long-running process has completed, but before the callers transaction is committed: If the caller was started by a persistent message, upon server restart the callers transaction is rolled back and retried. However, the result of the execution of the long-running process on the server is not rolled back, because it was committed before the server failure. As a result, the long-running process on the server is executed twice. This duplication causes functional problems in the application unless corrected manually. If the caller was not started by a persistent message, and the response of the long-running process was not submitted yet, it ends in the failed event queue.

Minimize number and size of BPEL variables and business objects

Follow these guidelines when defining BPEL variables and business objects: Use as few variables as possible and minimize the size and the number of business objects used. In long-running processes, each commit saves modified variables to the database (to save context), and multiple variables or large business objects make this costly. Smaller business objects are also more efficient to process when emitting monitor events. Specify variables as data type variables. This improves runtime performance. Use transformations (maps or assigns) to produce smaller business objects by only mapping fields necessary for the business logic.

2.3.4 Human task considerations

Adhere to these guidelines when developing human tasks for BPEL business processes: Use group work items for large groups (people assignment criterion: Group) instead of individual work items for group members (people assignment criterion: Group Members). Use native properties on the task object rather than custom properties where possible. For example, use the priority field instead of creating a new custom property priority. Set the transactional behavior to commit after if the task is not part of a page-flow. This improves the response time of task complete API calls.

2.3.5 Business process and human tasks client considerations

The following list details considerations for developing BPEL business process and human task clients: APIs that provide task details and process details, such as htm.getTask(), should not be called frequently. Use these methods only when required to display the task details of a single task, for instance. Do not put too much work into a single client transaction. In servlet applications, a global transaction is typically not available. If the servlet calls the HTM and BFM APIs directly, transaction size is typically not a concern.
Chapter 2. Development best practices

29

4784ch02.fm

Draft Document for Review January 27, 2012 7:40 am

In EJB applications, make sure that transactions are not too time consuming. Long-running transactions create long-lasting locks in the database, which prevent other applications and clients to continue processing. Chose the protocol that best suits your needs. In a J2EE environment, use the HTM and BFM EJB APIs. If the client application is running on a IBM BPM 7.5 Server, use the local EJB interface. In a Web 2.0 application, use the REST API. In an application that runs remote to the process container, the Web services API is an option. Clients that follow a page-flow pattern should consider the following issues: Use the completeAndClaimSuccessor() API if possible. This provides optimal response time. Applications that assign the next available task to the user can use the claim(String queryTableName, ) method on the Human Task Manger EJB interface. It implements a performance-optimized mechanism to handle claim collisions. Do not put asynchronous invocations between two steps of a page-flow, because the response time of asynchronous services increases as the load on the system increases. Where possible, do not invoke long-running sub-processes between two steps of a page-flow, because long-running sub-processes are invoked using asynchronous messaging. Clients that present task lists and process lists to the user should consider the following factors: Use query tables for task list and process list queries. Do not loop over the tasks displayed in the task or process list and execute an additional remote call for each object. This prevents the application from providing good response times and good scalability. Design the application so that all information is retrieved from a single query table during task list and process list retrieval. For instance, do not make calls to retrieve the input message for task list or process list creation.

2.3.6 Transactionality considerations

One of the strengths of the IBM BPM 7.5 platform is the precise control it provides for specifying transactional behavior. We recommend that when modeling a process or mediation assembly, the modeler should carefully design their desired transaction boundaries as dictated by the applications needs. Transaction boundaries are expensive in system resources. The objective of this section is to guide the modeler in avoiding unnecessary transaction boundaries. The following list details some guiding principles: The throughput of a particular usage scenario is inversely related to the number of transaction boundaries traversed in the scenario, so fewer transactions is faster. In user-driven scenarios, improving response time might require more granular transaction boundaries, even at the cost of throughput. Transactions can span across synchronous invocations, but cannot span asynchronous invocations. Avoid synchronous invocation of a two-way asynchronous target. The caller transactions failure recovery can be problematic.

30

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch02.fm

Exploit SCA transaction qualifiers

In an SCA assembly, the number of transaction boundaries can be reduced by allowing transactions to propagate across components. For any pair of components where this is desired, we recommend using the following settings: SuspendTransaction= false, for the calling components reference joinTransaction= true, for the called components interface Transaction= any|global, for both components implementation These settings assume that the first component in such a chain either starts or participates in a global transaction.

Avoid two-way synchronous invocation of an asynchronous target

If the target component has to be invoked asynchronously and its interface is of two-way request/response style, the target cannot be safely invoked through synchronous SCA calls. After the caller sends the request to the target, it waits for response from the target. Upon receiving the request, the asynchronous target starts a new transaction, and upon completion of the request processing returns the response asynchronously to the caller through the response queue. If a system failure occurs after the caller successfully sends the request but before receiving the response, the caller transaction is rolled back and retried. As a result, the target is invoked a second time.

Exploit transactional attributes for activities in long-running processes

While SCA qualifiers control component-level transactional behavior, there are additional transactional considerations in long-running business processes that can cause activities to be run in multiple transactions. The scope of those transactions and the number of transactions can be changed with the transactional behavior settings on Java Snippet, Human Task, and Invoke activities. For a detailed description of these settings, see the Information Center article Transactional behavior of business processes on the following website: http://publib.boulder.ibm.com/infocenter/dmndhelp/v7r0mx/index.jsp?topic=/com.ibm. websphere.bpc.doc/doc/bpc/cprocess_transaction.html There are four choices for transactional behavior settings: Commit before Commit after Participates Requires own Only the Participates setting does not require a new transaction boundary. The other three require the process flow container to start a new transaction before executing the activity, after executing the activity, or both. In general, the Participates attribute provides the best throughput and should be used wherever possible. This is true for both synchronous and asynchronous activities. In the two-way asynchronous case, it is important to understand that the calling transaction always commits after sending the request. The Participates setting refers to the transaction started by the process engine for the response. When set, this allows the next activity to continue on the same transaction. In special cases, the other transaction settings might be preferable. See the aforementioned Information Center article for details. Use Commit before in parallel activities that start new branches to ensure parallelism. As noted in the Information Center, there are other constraints to be considered.
Chapter 2. Development best practices

31

4784ch02.fm

Draft Document for Review January 27, 2012 7:40 am

Use Commit after for inline human tasks to increase responsiveness to human users. When a human user issues a task completion, the thread/transaction handling that action is used to resume navigation of the human task activity in the process flow. The users task completion action does not complete until the process engine commits the transaction. If the Participates setting is used, the commit gets delayed and forces a longer response time to the user. This is a classic response time versus throughput tradeoff. Starting with the 6.2.0 release, Receive and Pick activities in BPEL flow are allowed to define their own transactional behavior property values. If not set, the default value of initiating a Receive or Pick activity is Commit after. Consider using Participates where possible, because that behavior performs better.

2.3.7 Invocation style considerations

This section discusses invocation style considerations.

Use asynchrony judiciously

Components and modules might be wired to each other either synchronously or asynchronously. The choice of interaction style can have a profound impact on performance. Care should be exercised when making this choice.

Set the preferred interaction style to Synchronous whenever possible

Many IBM BPM 7.5 server component types (such as interface maps or business rules) invoke their target components based on the target interfaces Preferred Interaction Style setting. Because synchronous cross-component invocations are better performing, it is recommended to set the Preferred Interaction Style setting to Synchronous whenever possible. Only in specific cases (for example when invoking a long-running business process, or more generally whenever the target component requires asynchronous invocation) should this be set to Asynchronous. Starting with the 6.2 release of the IBM Integration Developer, when a new component is added to an assembly diagram, its Preferred Interaction Style is set to Synchronous, Asynchronous, or Any based on the component. In previous releases of the Integration Developer, the default initial setting of Preferred Interaction Style is set to Any unless explicitly changed by the user. If a components Preferred Interaction Style is set to Any, how the component is invoked is determined by the callers context. If the caller is a long running business process, a Preferred Interaction Style setting of Any is treated as asynchronous. If the caller is a non-interruptible business flow, a Preferred Interaction Style setting of Any is treated as synchronous. The invocation logic of processes is explained in more detail in the IBM Information Center at the following website: http://publib.boulder.ibm.com/infocenter/dmndhelp/v7r0mx/index.jsp?topic=/com.ibm. websphere.bpc.doc/doc/bpc/cprocess_transaction.html The following list details additional considerations: When setting an interfaces Preferred Interaction Style to Asynchronous, it is important to realize the downstream implications. Any components invoked downstream inherits the asynchronous interaction style unless they explicitly set the Preferred Interaction Style to Synchronous. At the input boundary to a module, exports that represent asynchronous transports such as MQ, JMS, or JCA (with asynchronous delivery set) set the interaction style to

32

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch02.fm

Asynchronous. This can cause downstream invocations to be asynchronous if the Preferred Interaction Style is Any. For an SCA import, its Preferred Interaction Style can be used to specify whether the cross-module call should be Synchronous or Asynchronous. For other imports that represent asynchronous transports such as MQ or JMS, it is not necessary to set the Preferred Interaction Style to Asynchronous. Doing so introduces an unnecessary asynchronous hop between the calling module and the invocation of the transport.

Minimize cross-component asynchronous invocations within a module

Asynchronous invocations are intended to provide a rich set of qualities of service, including transactions, persistence, and recoverability. Hence, an asynchronous invocation should be thought of as a full messaging hop to its target. When the intended target of the invocation is in the same module, a synchronous invocation yields better performance. Some qualities of services (such as event sequencing and store-and-forward) can only be associated with asynchronous SCA calls. Consider the performance impact of asynchronous invocations when setting these qualities of service.

Avoid async invocation of sync services in a FanOut / FanIn block

Do not select asynchronous (deferred response interaction) service invocations for services with synchronous bindings (for example, Web services) unless there is an overriding need for this, and the non-performance implications for this style of invocation are understood. Apart from the performance implications of calling a synchronous service asynchronously, there are reliability and transactional aspects to be considered. Make sure you understand the non-performance implications of using asynchronous callouts before considering their use. Generally, asynchronous callouts should only be used for idempotent query type services. If you need to guarantee that the service is only called once, do not use asynchronous invocation. It is beyond the scope of this paper to provide complete guidance on the functional applicability of using asynchronous callouts in your mediation flow. More information can be found in the IBM Integration Developer help documentation and IBM BPM 7.5 Information Centers. The Information Centers for all three products can be found at the following website: http://publib.boulder.ibm.com/infocenter/dmndhelp/v7r5mx/index.jsp Assuming that asynchronous callouts are functionally applicable for you there might be a performance reason for invoking a service in this style but it should be understood that asynchronous processing is inherently more expensive in terms of CPU cycles due to the additional messaging overhead incurred by calling a service this way. There are additional operational considerations. For example, asynchronous invocations use the service integration bus messaging infrastructure, which uses a database for persistence. Synchronous invocations perform well with basic tuning of the JVM heap size and thread pools but for asynchronous invocations, SCA artifacts require review and tuning. This includes tuning of the SCA messaging engine (see 3.3.7, Messaging engine properties on page 46), data sources (3.3.6, JDBC data source parameters on page 45) and the database itself. For the data source, the tunings for JMS bindings in this paper can be used as guidance as the considerations are the same. If multiple synchronous services with large latencies are being called, asynchronous invocations can reduce the overall response time of the mediation flow at the expense of increasing the internal response time of each individual service call. This assumes that

Chapter 2. Development best practices

33

4784ch02.fm

Draft Document for Review January 27, 2012 7:40 am

asynchronous callouts have been configured along with parallel waiting in the FanOut section of the flow: In the case of iteration of array, configuring the FanOut to check for asynchronous responses after all/N messages have been fired. In case of extra wires/FlowOrder primitive - by default. If there are a number of services in a fan-out section of a mediation flow, calling these synchronously results in an overall response time equal to the sum of the individual service response times. Calling the services asynchronously (with parallel waiting configured) results in a response time equal to at least the largest individual service response time in WebSphere ESB plus the sum of the time taken by WebSphere ESB to process the remaining service callout responses residing on the messaging engine queue. For a FanOut/FanIn block the processing time for any primitives before or after the service invocations needs to be added in both cases. To optimize the overall response time when calling services asynchronously in a FanOut/FanIn section of a mediation flow, invoke the services in the order of expected latency (highest latency first), if known. There is a trade off between parallelism and additional asynchronous processing to consider. The suitability of asynchronous processing depends on the size of the messages being processed, the latency of the target services, the number of services being invoked, and any response time requirements expressed in service level agreements. Running performance evaluations on mediation flows including fan-outs with high latency services is recommended if asynchronous invocations are being considered. The default quality of service on service references is Assured Persistent. A substantial reduction in asynchronous processing time can be gained by changing this to Best Effort (non-persistent), which eliminates I/O to the persistence store but the application must tolerate the possibility of lost request or response messages. This level of reliability for the service integration bus can discard messages under load and might require tuning.

2.3.8 Mediation flow considerations

This section discusses mediation flow considerations for performance.

Usage of XSLTs versus business object maps

The XSLT and the Business Object Map primitives offer two alternative transformation approaches in a mediation flow. If no XSLT specific function is required then it is generally better to use the Business Object Map primitive in terms of performance which can be significantly faster. The exception is when a mediation flow component is trivial in nature and the transformation is taking place at the /body level of the SMO. In this case XSLT will be faster as the native XML bytes can be passed straight to the XSLT engine. Native XML bytes will be available if the XSLT transform primitive is the first in the flow or only preceded by one or more of the following primitives: Route on Message Header (Message Filter Primitive) XSLT Primitive (Transforming on /body as the root) EndpointLookup without Xpath user properties. 34
IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch02.fm

Event Emitter (CBE Header Only)

Aggregation Blocks (FanOut/FanIn)

Aggregation is an important Websphere ESB pattern. It enables a single inbound request to map into multiple outbound service invocations, the responses from which can be aggregated into a single response to the original request. Before you develop mediations using aggregation design patterns you need to take several performance considerations into account. A developerWorks article on best practises is available at: http://www.ibm.com/developerworks/websphere/library/techarticles/1111_norris/1111_ norris.html

Configure WebSphere ESB resources

When creating resources using IBM Integration Developer, the application developer is given the choice to use pre-configured WebSphere ESB resources or to let the tooling generate the mediation flow-related resources that it requires. Both approaches have their advantages and disadvantages. Pre-configured resources support the following circumstances: Existing resources to be used External creation and tuning scripts to be applied Easier post deployment adjustment Tooling-created resources support the following circumstances: No further need for creating resources using scripts or the Admin Console. The ability to change the majority of performance tuning options as they are now exposed in the Tooling. In our performance tests we use pre-configured resources. This is because by segregating the performance tuning from the business logic, the configuration for different scenarios can be maintained in a single script. It is also easier to adjust these parameters after the applications have been deployed. The only cases where this pattern was not followed in our tests was for generic JMS bindings. In scenarios where resources have already been configured by the third-party JMS provider software (MQ 6.0.2.2 for all instances in this paper), the tooling-created resources were used to locate the externally defined resources.

2.4 WebSphere InterChange Server migration considerations

The following list contains useful information for those migrating from WebSphere InterChange Server to IBM BPM 7.5.0: Utilize WebSphere Adapters to replace WebSphere Business Integration adapters (WBI adapters), where possible. Migrated workloads making use of custom WBI adapters or legacy WBI adapters result in interaction with IBM BPM 7.5 through JMS, which is slower than the JCA adapters. Some technology adapters (such as HTTP and Web services) are migrated by the WebSphere InterChange Server migration wizard into native IBM BPM 7.5 SCA binding, which is a better performing approach. For adapters that are not migrated automatically to

Chapter 2. Development best practices

35

4784ch02.fm

Draft Document for Review January 27, 2012 7:40 am

available SCA bindings, development effort spent to migrate manually to an SCA binding removes the dependency on a legacy adapter and provides better performance. The WebSphere InterChange Server Migration Wizard in IBM Integration Developer offers a feature to merge the connector and collaboration module together. Enable this option, if possible, as it increases performance by reducing cross-module SCA calls. WebSphere InterChange Server Collaborations are migrated into IBM BPM 7.5 BPEL processes. The resultant BPEL processes can be further customized and made more efficient. Migrated BPEL processes enable support for compensation by default. If the migrated workload does not make use of compensation, this support can be disabled to gain performance. The relevant flag can be found in the IBM Integration Developer by navigating to process name properties Details Require a compensation sphere context to be passed in. The generated BPEL flows still make use of the ICS APIs to perform business object and collaboration level tasks. Development effort spent cleaning up the migrated BPEL to use IBM BPM APIs instead of the ICS APIs results in better performance and better maintainability.

Investigate the possibility of replacing BPEL processes produced by migration with other artifacts. All WebSphere InterChange Server collaborations currently get migrated into BPEL processes. For certain scenarios, other IBM BPM 7.5 server artifacts (for example business rules) might be better choices. Investigate the BPEL processes produced by migration to ensure the processes are the best fit for your scenario. Disable Message Logger calls in migration-generated mediation flow component (MFC) components. The WebSphere InterChange Server Migration Wizard in IBM Integration Developer 7.5 generates an MFC to deal with the mapping details of a connector. It contains the code for handling synchronous and asynchronous calls to maps that transform application-specific business objects to generic business objects and generic business objects to application-specific objects. The generated MFC contains embedded MessageLogger calls that log the message to a database. Disable these calls if not required in your business scenario to reduce writes to the database and thus improve performance. (Select the MessageLogger instance, choose the Details panel, and clear the Enabled checkbox.) Reduce memory pressure by splitting the shared library generated by the migration wizard. The migration wizard creates a single shared library and puts all migrated business objects, maps, and relationships in it. This library is then shared by copy by all the migrated modules. This can cause memory bloat for cases where the shared library is large and a large number of modules are present. The solution is to re-factor manually the shared library into multiple libraries based on functionality or usage and modify modules to only reference the shared libraries that they need. If the original WebSphere InterChange Server maps contain many custom map steps, the development effort spent in rewriting those map steps results in better performance. The WebSphere InterChange Server Migration Wizard in IBM Integration Developer 7.5 generates maps that make use of the ICS APIs, which is a translation layer above IBM BPM 7.5 server technologies. Removing this layer by making direct use of IBM BPM 7.5 Server APIs avoids the cost of translation and produces better performance.

36

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch02.fm

2.5 Authoring environment considerations

This section describes recommendations intended to improve the performance of activities commonly encountered by application developers during the development of an enterprise application, primarily the import, build, and publish of an application workspace. This is in contrast to the previous sections focus on using the authoring tools to produce high performance runtime applications.

Use hardware advantages

Importing and building an enterprise application is a resource intensive activity. Recent improvements in desktop hardware architecture have greatly improved the responsiveness of import and build activities. Also, for I/O intensive activities (such as Import) a faster disk drive or disk subsystem reduces total response time.

Share Business Integration libraries across modules

BPM 7.5 adds the IBM Process Center (IPC), which allows Business Integration modules to become part of a BPM Process App, which is a new concept in this release. One of the benefits of a Process App is the ability to scope Business Integration libraries to be shared among the various modules in a Process App instead of being duplicated for each module. This allows server memory to be saved for large libraries that were previously module scoped, since duplicate copies can be eliminated. Lazy Business Object parsing mode is required to use the Process App scope. While the Process App scope is the default for new libraries in 7.5 Process Apps, existing libraries retain the Module scope. Whenever possible, these should be changed to use the Process App scope. Memory savings are correlated to the number of modules that reference the library. The greater the number of modules that use the library, the greater the potential memory savings for using Process App scoped libraries.

Place libraries in Process App Scope to save memory

New in BPM 7.5.0 is the IBM Process Center (IPC). This allows Business Integration modules to become part of a BPM Process App, which is a new concept in this release. One of the benefits of a Process App is the ability to scope Business Integration libraries to be shared among the various modules in a Process App instead of being duplicated for each module. This allows server memory to be saved for large libraries that were previously module scoped, since duplicate copies can be eliminated. In v7.5.0, Lazy BO parsing mode is required to use the Process App scope. While the Process App scope is the default for new libraries in v7.5.0 Process Apps, existing libraries retain the Module scope. Whenever possible, these should be changed to use the Process App scope. Memory savings are correlated to the number of modules that reference the library. The greater the number of modules that use the library, the greater the potential memory savings for using Process App scoped libraries.

2.6 Business Space Developer Considerations

While developing a Business Space (BSpace) solution, developers will often alter the configuration of the browser for debugging, tracing, etc. However, some of these configuration changes can dramatically affect the performance of the BSpace solution. Before rolling out a solution to production, or taking performance measurements, please review this section and adhere to these guidelines.

Chapter 2. Development best practices

37

4784ch02.fm

Draft Document for Review January 27, 2012 7:40 am

Do not use Addins or extensions for production use

An important browser consideration is the use of addins/extensions for both Internet Explorer (IE) and Firefox. For IE, having the Developer Tools enabled dramatically decreases performance. To some extent, Firebug and HttpWatch do as well, but not to the same degree. An end user who is experiencing problems should always make sure that they have all addins/extensions disabled or un-installed unless they are absolutely necessary.

IE: Dont use Developer Tools Render/Compatibility Mode for production

In IE it is possible to place the browser in compatibility mode, or in a render mode equivalent to an older browser version, by using the Developer Tools. While this can be useful for developing or debugging solutions (e.g. to have an IE 8 browser behave as an IE 7 browser for compatibility testing), end users should make sure they do not utilize these settings, since more recent versions of IE typically perform better than older versions.

Browser Tools
Following are observations for a few of the browser tools that are available for obtaining response time measurements, counting requests, and analyzing caching. Firebug's Net tab is very useful for obtaining request timings, analyzing request/response headers, and counting the number of requests. However it reports requests that are satisfied from the browser cache as 200 responses. You can still determine that it is a cached request because there will be a "Cache" tab on the request which will report that the request was actually satisfied from the browser cache. However, if you copy and paste the results into another document (i.e an e-mail note) the cache tab is not copied, so it is possible to be misled by the 200 responses and draw an incorrect conclusion that caching isn't effective when it actually is. Fiddler is another extremely powerful tool, which also has the advantage of supporting both IE and FireFox browsers. However, since it acts as a proxy between the browser and the server and cached requests are handled internally by browsers, they never appear in Fiddler. This clearly is detrimental in determining which requests are fulfilled from the browser cache, but on the other hand it is useful to use Fiddler to analyze only the requests that actually are sent to the server. HttpWatch does not have the limitations described above, since it is supported on both IE and FireFox browsers, the results copy and paste easily both into a spreadsheet or a document, and it displays cached requests in a straightforward manner.

38

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch03.fm

Chapter 3.

Performance tuning and configuration

To optimize performance it is usually necessary to configure the system differently than the default settings. This chapter lists several areas to consider during system tuning. This includes tuning the IBM BPM products, and other products in the system, like DB2. The documentation for each of these products contains a wealth of information regarding performance, capacity planning, and configuration. This documentation would likely offer the best guidance for performance considerations in a variety of operational environments. Assuming that all these issues have been addressed from the perspective of the actual product, additional levels of performance implications are introduced at the interface between these products and the products covered in this paper. A number of configuration parameters are available to the system administrator. While this chapter identifies several specific parameters observed to affect performance, it does not address all available parameters. For a complete list of configuration parameters and possible settings see the relevant product documentation. The first section of this chapter describes a methodology to use when tuning a deployed system. It is followed by a basic tuning checklist that enumerates the major components and their associated tuning concepts. The subsections that follow address tuning in more detail, describing several tuning parameters and their suggested setting (where appropriate), and providing advanced tuning guidelines for more detailed guidance for key areas of the system. Representative values for many of the tuning parameters are shown in Chapter 4, Initial configuration settings on page 81. Important: While there is no guarantee that following the guidance in this chapter will provide acceptable performance, it is likely that degraded performance can be expected if these parameters are incorrectly set. Related publications on page 89 contains references to related documentation that might prove valuable when tuning a particular configuration.

Copyright IBM Corp. 2012. All rights reserved.

39

4784ch03.fm

Draft Document for Review January 27, 2012 7:40 am

3.1 Performance tuning methodology

We recommend a system-wide approach to performance tuning of IBM BPM environment. System performance tuning, which requires training and experience, is not going to be exhaustively described here. Rather, we highlight some key aspects of tuning that are particularly important. Tuning encompasses every element of the deployment topology: Physical hardware topology choices Operating System parameters tuning IBM BPM Server, WebSphere Application Server, and messaging engine tuning The methodology for tuning can be stated as an iterative loop: 1. 2. 3. 4. Pick a set of reasonable initial parameter settings and run the system. Monitor the system to obtain metrics that indicate whether performance is being limited. Use monitoring data to guide further tuning changes. Repeat until done.

The following subsections examine these steps.

3.1.1 Pick a set of reasonable initial parameter settings

Use the tuning checklist in 3.2, Tuning checklist on page 41 for a systematic way to set parameters. For specific initial values, consult Chapter 4, Initial configuration settings on page 81 for settings that were used for the various workloads used for our internal performance evaluation. These values can be considered for initial values.

3.1.2 Monitor the system

We recommend monitoring the systems to determine system health, as well as to determine the need for further tuning. The following should be monitored: For each physical machine in the topology including front-end and back-end servers such as Web servers, and database servers, monitor the following processes using relevant OS tools (such as vmstat, iostat, netstat, or equivalent): Processor core utilization Memory utilization Disk utilization Network utilization

For each JVM process started on a physical machine (for example, process server, messaging engine server, and so forth): Use tools such as ps or equivalent to get core and memory usage per process. Collect verbosegc statistics. For each IBM BPM Server or messaging engine JVM, use Tivoli Performance Viewer to monitor: Data connection pool utilization for each data source Thread pool utilization for each thread pool (Web container, default, work managers)

40

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch03.fm

3.1.3 Use monitoring data to guide further tuning changes

This is a topic that requires skill and experience. In general, this phase of tuning requires the analyst to look at the collected monitoring data, detect performance bottlenecks, and do further tuning. The key characteristic about this phase of tuning is that it is driven by the monitoring data collected in the previous phase. Examples of performance bottlenecks include, but are not limited to: Excessive utilization of physical resources such as processor cores, disk, memory, and so forth. These can be resolved either by adding more physical resources, or rebalancing the load across the available resources. Excessive utilization of virtual resources. Examples include heap memory, connection pools, thread pools, and so forth. For these, tuning parameters should be used to remove the bottlenecks.

3.2 Tuning checklist

This checklist serves as a guide when tuning a IBM BPM solution. Each of these topics is covered in more detail in the remainder of this chapter. Common Utilize a 64-bit JVM for all servers Disable tracing and monitoring when possible. Ensure all databases reside on a high performance DBMS such as DB2, and that the databases are well-tuned If security is required, use Application security, not Java2 security. Use an appropriate hardware configuration for performance measurement (for example, laptops and desktops are not appropriate for realistic performance evaluations). If hardware virtualization is used, ensure that adequate processor, memory, and I/O resources are allocated to each virtual machine. Avoid over-committing resources. Do not run production servers in development mode or with a development profile. Tune external service providers and external interfaces to ensure they are not the system bottleneck. Configure message-driven bean activation specifications Configure for clustering (where applicable) Configure thread pool sizes Configure data sources settings for connection pool size and prepared statement cache size. Consider using non-XA data sources for CEI data when that data is non-critical. Increase the maximum number of connections in the data pool to greather than or equal to the sum of all maximum thread pool sizes.

Business Process Choreographer (for BPEL business processes) Use work-manager based navigation for long running processes, and optimize the message pool size and intertransaction cache size. Use query tables to optimize query response time

Chapter 3. Performance tuning and configuration

41

4784ch03.fm

Draft Document for Review January 27, 2012 7:40 am

Optimize Business Flow Manager resources: Database connection (BPEDB), Activation specification (BPEInternalActivationSpec) JMS connection (BPECF and BPECFC)

Optimize the database configuration for the Business Process Choreographer database (BPEDB) Optimize indexes for SQL statements that result from task and process list queries using database tools such as the DB2 design advisor Turn off state observers that are not needed (for example, turn off audit logging). BPMN Business Processes Increase log file size for twproc database (to 16384 pages) Enable filesystem caching for twproc database: db2 alter tablespace userspace1 file system caching Exclude the table, SIBOWNER, from automatic runstats execution Ensure that database statistics are up to date. Create new indexes as described in the technote at: http://www-01.ibm.com/support/docview.wss?uid=swg21474536 Messaging and message bindings Optimize activation specification (JMS, MQJMS, MQ) Optimize queue connection factory (JMS, MQJMS, MQ) Configure connection pool size (JMS, MQJMS, MQ) Configure service integration bus data buffer Sizes Place database tablespaces and logs on a fast disk subsystem Place logs on a separate device from the tablespace containers Maintain current indexes on tables Update database statistics Set log file sizes correctly Optimize buffer pool size (DB2) or buffer cache size (Oracle)

Database

Java Set the heap and nursery sizes to manage memory efficiently Choose the appropriate garbage collection policy (generally, -Xgcpolicy:gencon) Business Monitor Configure CEI Set message consumption batch size Enable KPI caching Use table-based event delivery Enable the data movement service

3.3 Common tuning parameters

This section lists performance tuning parameters commonly used for tuning IBM BPM 7.5 solutions, for both BPMN and BPEL business processes.

42

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch03.fm

3.3.1 Tracing and logging flags

Tracing and logging are often necessary when setting up a system or debugging issues. However, these capabilities produce performance overhead that is often significant. Minimize their use when evaluating performance or in production environments. This section lists tracing parameters used in the products covered in this paper. Some settings are common to all or a subset of the products, while others are specific to a particular product. Unless stated otherwise, all of these parameters can be set using the administrative console. To enable or disable tracing, go to Troubleshooting Logs and Trace server_name Change Log Detail Levels and set both the Configuration and Runtime to *=all=disabled. To change the PMI level go to Monitoring and Tuning Performance Monitoring Infrastructure server_name and select none. In addition, Cross-Component Tracing (XCT) is useful for problem determination, enabling correlation of SCA component information with log entries. However, XCT should not be used in production or while obtaining performance data. There are two levels of XCT settings: Enable Enable with data snapshot Both incur significant performance overhead. Enable with data snapshot is particularly costly because of the additional I/O involved in saving snapshots in files. To enable or disable XCT, go to Troubleshooting Cross-Component Trace. Select the XCT setting from three options in the Configuration or Runtime tab: Disable Enable Enable with data snapshot Changes made on the Runtime tab take effect immediately. Changes made on the Configuration tab require a server restart to take effect. Further information is provided in the technote available at: http://www-01.ibm.com/support/docview.wss?uid=swg21439659

3.3.2 Java tuning parameters

In this section we list a few frequently used Java Virtual Machine (JVM) tuning parameters. For a complete list, consult the JVM tuning guide offered by the JVM supplier. The JVM administrative panel can be accessed from Servers Application Servers server_name Server Infrastructure Java and Process Management Process Definition Additional Properties Java Virtual Machine.

Java garbage collection (GC) policy

The default garbage collection algorithm on platforms with an IBM JVM is a generational concurrent collector (specified with -Xgcpolicy:gencon under the Generic JVM arguments on the JVM administrative console panel). Our internal evaluation shows that this garbage collection policy usually delivers better performance with a tuned nursery size as discussed in the next section.

Chapter 3. Performance tuning and configuration

43

4784ch03.fm

Draft Document for Review January 27, 2012 7:40 am

Java heap sizes

To change the default Java heap sizes, set the initial heap size and maximum heap size explicitly on the JVM panel in the administrative console. Note that 64-bit JVMs (the recommended mode for IBM BPM 7.5 servers) support much larger heap sizes than 32-bit JVMs. Uitilze this capability to relieve memory pressure in the Java heap, but always ensure that there is sufficient physical memory to back the JVM heap size and all other memory requirements. If Generational Concurrent Garbage Collector is selected, the Java heap is divided into a new area (nursery) where new objects are allocated and an old area (tenured space) where longer lived objects reside. The total heap size is the sum of the new area and the tenured space. The new area size can be set independently from the total heap size. Typically the new area size should be set between 1/4 and 1/2 of the total heap size. The relevant parameters are: Xmns<size> : initial new area size Xmnx<size> : maximum new area size Xmn<size> : fixed new area size

3.3.3 MDB ActivationSpec

There are a few shortcuts you can take in the administrative console to access the MDB ActivationSpec tuning parameters: Resources Resource Adapters J2C activation specifications ActivationSpec_name Resources JMS Activation specifications ActivationSpec_name Resources Resource Adapters Resource adapters resource_adapter_name Additional properties J2C activation specifications ActivationSpec_name The following two custom properties in the MDB ActivationSpec have considerable performance implications. These are discussed further in Tune MDB ActivationSpec properties on page 48. maxConcurrency maxBatchSize

3.3.4 Thread pool sizes

IBM BPM Servers use thread pools to manage concurrent tasks. The Maximum Size property of a thread pool can be set in the administrative console by going to Servers Application servers server_name Additional Properties Thread Pools thread_pool_name. The following thread pools typically need to be tuned: Default ORB.thread.pool WebContainer In addition, thread pools used by Work Managers for BPEL processes are configured separately in the console by going to Resources Asynchronous beans Work managers work_manager_name Thread pool properties The following Work Managers typically need to be tuned: 44
IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch03.fm

DefaultWorkManager BPENavigationWorkManager

3.3.5 JMS connection pool sizes

There a few ways of accessing the JMS connection factories and JMS queue connection factories from the administrative console.: Resources Resource Adapters J2C connection factories factory_name Resources JMS Connection factories factory_name Resources JMS Queue connection factories factory_name Resources Resource Adapters Resource adapters resource_adapter_name (for example, SIB JMS Resource Adapter) Additional priorities J2C connection factories factory_name From the connection factory admin panel, open Additional Properties Connection pool properties. Set the Maximum connections property for the maximum size of the connection pool.

3.3.6 JDBC data source parameters

Data sources can be accessed through either of these paths: Resources JDBC Data sources data_source_name Resources JDBC Providers JDBC_provider_name Additional Properties Data sources data_source_name

Connection pool size

The maximum size of the data source connection pool is limited by the value of the Maximum connections property, which can be configured from the data source panels Additional Properties Connection pool properties. Increase the maximum number of connections in the data pool to greather than or equal to the sum of all maximum thread pool sizes. The following data sources typically need to be tuned: BPC data sources for the BPE database and associated message engine database (for BPEL business processes) BPMN data sources for the BPMN databases and associated message engine databases (for BPMN business processes) SCA application bus messaging engine data source SCA system bus messaging engine data source CEI bus messaging engine data source

Prepared statement cache size

The data source prepared statement cache size can be configured from the data sources Additional properties WebSphere Application Server data source properties. For IBM BPM Servers, the BPEDB data source should typically be tuned to a higher value. Start by setting this to 300.

Chapter 3. Performance tuning and configuration

45

4784ch03.fm

Draft Document for Review January 27, 2012 7:40 am

3.3.7 Messaging engine properties

Two messaging engine custom properties might impact the messaging engine performance: sib.msgstore.discardableDataBufferSize In memory buffer for best effort nonpersistent messages. Default is 320 K. Once full, messages are discarded to allow newer messages to be written to the buffer. sib.msgstore.cachedDataBufferSizeCachedDataBufferSize In memory cache for messages other than best effort nonpersistent Default is 320 K These properties can be accessed in the console by selecting Service Integration Buses bus_name Messaging Engines messaging_engine_name Additional properties Custom properties. Full details of these are given in the Information Center article at the following website: http://publib.boulder.ibm.com/infocenter/wasinfo/v7r0/index.jsp?topic=/com.ibm.web sphere.pmc.doc/concepts/cjk_learning.html

3.3.8 Run production servers in production mode

IBM BPM servers can be run in development mode to reduce startup time for the server by using JVM settings to disable bytecode verification and reduce JIT compilation time. This setting should not be used on production servers, because it is not designed to produce optimal runtime performance. Make sure the Run in development mode checkbox for the server is cleared. This setting is found in the servers configuration panel in the administrative console. Go to Servers Application Servers server_name Configuration. Server profiles might also be created with production or development templates. Use production profile templates for production servers.

3.4 Advanced tuning

This section contains advanced tuning tips.

3.4.1 Tracing and monitoring considerations

The ability to configure tracing and monitoring at different levels for a variety of system components has proven to be extremely valuable during periods of system analysis or debugging. The IBM BPM product set provides rich monitoring capabilities, both in terms of business monitoring through the Common Event Interface (CEI) and audit logging, and system performance monitoring through the Performance Monitoring Infrastructure (PMI) and the Application Response Measurement (ARM) infrastructure. While these capabilities provide insight into the performance of the running solution, these features can degrade overall system performance and throughput. Note: Use tracing and monitoring judiciously. When possible, turn them off to ensure optimal performance.

46

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch03.fm

Most tracing and monitoring is controlled through the administrative console. Validate that the appropriate level of tracing and monitoring is set for the PMI Monitoring, Logging, and Tracing settings through the administrative console. Use the administrative console to validate that the Audit logging and Common Event Infrastructure logging check boxes are cleared in the Business Flow Manager and the Human Task Manager, unless these capabilities are required for business reasons. IBm Integration Developer is also used to control event monitoring. Check the Event Monitor tab for your components and business processes to ensure that event monitoring is applied judiciously.

3.4.2 Tuning for large objects

In this section we discuss tuning for performance when using large objects.

Heap limitations: Increase the Java heap to its maximum

One of the key factors affecting large object processing is the maximum size of the Java heap. In this section we discuss how to set the heap size as large as possible on two commonly used platforms. Note that this only applies to 32-bit JVMs; when using 64-bit JVMs, the heap size is only constrained by the amount of physical memory available 64-bit is the recommended configuration for BPM 7.5 servers. For more comprehensive heap setting techniques, consult 3.13, Advanced Java heap tuning on page 76. Windows (32-bit) Due to address space limitations in the Windows 32-bit operating system, the largest heap that can be obtained is 1.4 GB1.6 GB for 32-bit JVMs. AIX (32-bit) On 32-bit systems, the Java5 and Java 6 JVM typically supports heaps in the range of 2 GB to 2.4 GB. Because the 4 GB address space allowed by the 32-bit system is shared with other resources, the actual limit of the heap size depends on memory usage by resources (such as thread stacks, JIT compiled code, loaded classes, shared libraries, buffers used by OS system services, and so forth). An extremely large heap squeezes address space reserved for other resources and might cause runtime failures.

Reduce or eliminate other processing while processing a large object

One way to allow for larger object sizes is to limit the concurrent processing within the JVM. Do not expect to process a steady stream of the largest objects possible concurrently with other IBM BPM Server and WebSphere Adapters activities. The operational assumption that needs to be made when considering large objects is that not all objects are large or very large and that large objects will not arrive often, perhaps once or twice per day. If more than one very large object is being processed concurrently the likelihood of failure increases dramatically. The size and number of the normally arriving smaller objects affects the amount of Java heap memory consumption in the system. Generally speaking, the heavier the load on a system when a large object is being processed, the more likely that memory problems are encountered. For adapters, the amount of concurrent processing can be influenced by setting the pollPeriod and pollQuantity parameters. To allow for larger object sizes, set a relatively high value for pollPeriod (for example, 10 seconds) and low value for pollQuantity (for example, 1) to
Chapter 3. Performance tuning and configuration

47

4784ch03.fm

Draft Document for Review January 27, 2012 7:40 am

minimize the amount of concurrent processing that occurs. These settings are not optimal for peak throughput, so if a given adapter instance must support both high throughput for smaller objects interspersed with occasional large objects, trade-offs must be made.

3.4.3 Tuning for maximum concurrency

For most high volume deployments on server-class hardware, there are many operations that take place simultaneously. Tuning for maximum concurrency ensures that the server accepts enough load to saturate its core. One sign of an inadequately-tuned configuration is when additional load does not result in additional core utilization while the cores are not fully utilized. To optimize these operations for maximum concurrency, the general guideline is to follow the execution flow and remove bottlenecks one at a time. Higher concurrent processing means higher resource requirements (memory and number of threads) on the server. It needs to be balanced with other tuning objectives, such as the handling of large objects, handling large numbers of users, and providing good response time.

Tune edge components for concurrency

The first step is to ensure that business objects are handled concurrently at the edge components of IBM BPM solutions. If the input business objects come from the adapter, ensure the adapter is tuned for concurrent delivery of input messages. If the input business objects come from WebServices export binding or direct invocation from JSPs or servlets, make sure the WebContainer thread pool is sized right. For example, to allow for 100 in-flight requests to be handled concurrently by an IBM BPM Server, the maximum size of the WebContainer thread pool needs to be set to 100 or larger. If the input business objects come from messaging, the ActivationSpec (MDB bindings) and Listener ports (MQ or MQJMS bindings) need to be tuned.

Tune MDB ActivationSpec properties

For each JMS export component, there is a message-driven bean (MDB) and its corresponding ActivationSpec (JNDI name: module name/export component name_AS). The default value for maxConcurrency of the JMS export MDB is 10, meaning up to 10 business objects from the JMS queue can be delivered to the MDB threads concurrently. Change it to 100 if a concurrency of 100 is desired. The Tivoli Performance Viewer can be used to monitor the maxConcurrency parameter. For each message being processed by an MDB there is a message on the queue marked as being locked inside a transaction (this is removed once the onMessage completes). These messages are classed as unavailable. There is a PMI metric called UnavailableMessageCount that gives you the number of unavailable messages on each queue point (select resource_name SIB Service SIB Messaging Engines bus_name Destinations Queues). If any queue has at least maxConcurrency unavailable messages, this would imply that the number of messages on the queue is currently running above the MDB's concurrency maximum. If this occurs, increase the maxConcurrency setting for that MDB. The maximum batch size in the activation specification also has an impact on performance. The default value is 1. The maximum batch size value determines how many messages are taken from the messaging layer and delivered to the application layer in a single step. (This does not mean that this work is done within a single transaction, and therefore this setting does not influence transactional scope) Increase this value (for example to 8) for activation

48

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch03.fm

specs associated with SCA modules and long-running business processes to improve performance and scalability, especially for large multi-core systems.

Configure thread pool sizes

The sizes of thread pools have a direct impact on a servers ability to run applications concurrently. For maximum concurrency, the thread pool sizes need to be set to optimal values. Increasing the maxConcurrency or Maximum sessions parameters only enables the concurrent delivery of business objects from the JMS or MQ queues. For an IBM BPM server to process multiple requests concurrently, it is necessary to increase the corresponding thread pool sizes to allow higher concurrent execution of these MDB threads. MDB work is dispatched to threads allocated from the default thread pool. All MDBs in the application server share this thread pool, unless a different thread pool is specified. This means that the Default thread pool size needs to be larger, probably significantly larger, than the maxConcurrency of any individual MDB. Threads in the WebContainer thread pool are used for handling incoming HTTP and Web Services requests. This thread pool is shared by all applications deployed on the server and it needs to be tuned, likely to a higher value than the default. ORB thread pool threads are employed for running ORB requests (for example, remote EJB calls). The thread pool size needs to be large enough to handle requests coming through the EJB interface, such as certain human task manager APIs.

Configure dedicated thread pools for MDBs

The Default thread pool is shared by many WebSphere Application Server tasks. It is sometimes desirable to separate the execution of JMS MDBs to a dedicated thread pool. Perform the following steps to change the thread pool used for JMS MDB threads. 1. Create a new thread pool (for example, MDBThreadPool) on the server by going to Servers Server Types WebSphere application servers server Thread pools. Click New. 2. Open the service integration bus (SIB) JMS Resource Adapter administrative console panel with server scope by going to Resources Resource Adapters Resource adapters. If the adapter is not shown, go to Preferences, and select the Show built-in resources checkbox. 3. Change the thread pool alias from Default to MDBThreadPool. 4. Repeat steps 2 and 3 for SIB JMS Resource Adapters at the node and cell scope. 5. Restart the server for the changes to be effective. SCA Module MDBs for asynchronous SCA calls use a separate resource adapter: the Platform Messaging Component SPI Resource Adapter. Follow the same steps to change the thread pool to a different one, if so desired. Even with a dedicated thread pool, all MDBs associated with the resource adapter still share the same thread pool. However, they do not have to compete with other WebSphere Application Server tasks that also use the Default thread pool.

Configure JMS and JMS queue connection factories

Multiple concurrently running threads might bottleneck on resources such as JMS and database connection pools if such resources are not tuned properly. The Maximum Connections pool size specifies the maximum number of physical connections that can be created in this pool. These are the physical connections to the back-end resource (for example, a DB2 database). After the thread pool limit is reached, no new physical
Chapter 3. Performance tuning and configuration

49

4784ch03.fm

Draft Document for Review January 27, 2012 7:40 am

connections can be created and the requester waits until a physical connection that is currently in use is returned to the pool, or a ConnectionWaitTimeout exception is issued. For example, if the Maximum Connections value is set to 5, and there are five physical connections in use, the pool manager waits for the amount of time specified in Connection Timeout for a physical connection to become free. The threads waiting for connections to the underlying resource are blocked until the connections are freed up and allocated to them by the pool manager. If no connection is freed in the specified interval, a ConnectionWaitTimeout exception is issued. If Maximum Connections is set to 0, the connection pool is allowed to grow infinitely. This has the side effect of causing the Connection Timeout value to be ignored. The general guideline for tuning connection factories is that their maximum connection pool size needs to match the number of concurrent threads multiplied by the number of simultaneous connections per thread. For each JMS, MQ, or MQJMS Import, there is a connection factory created during application deployment. The maximum connections property of the JMS connection factorys connection pool should be large enough to provide connections for all threads concurrently executing in the import component. For example, if 100 threads are expected to run in a given module, the maximum connections property should be set to 100. The default is 10. From the connection factory configuration panel, open Additional Properties Connection pool properties. Set the Maximum Connections property to the maximum size of the connection pool.

Configure data source options

The Maximum Connections property of data sources should be large enough to allow concurrent access to the databases from all threads. Typically there are a number of data sources configured in IBM BPM servers (for example, the BPEDB data source, the twproc data sources, the twperfdb data sources, the WPSDB data source, and the messaging engine database data sources). Set each data sources Maximum Connections property to match the maximum concurrency of other system resources as discussed in 3.4.3, Tuning for maximum concurrency on page 48.

Set data source prepared statement cache size

The BPC container uses prepared statements extensively. The statement cache sizes should be large enough to avoid repeatedly preparing statements for accessing the databases. The prepared statement cache for the BPEDB data source should be at least 300.

3.4.4 Messaging tuning

This section discusses performance tuning for messaging.

For messaging engines, choose data store or file store

Messaging engine persistence is usually backed by a database. However, a standalone IBM BPM configuration can have the persistence storage of BPE and SCA buses backed by the file system (file store). The choice of file store has to be made at profile creation time. Use the Profile Management Tool to create a new standalone enterprise service bus profile or standalone process server profile. Choose Profile Creation Options Advanced profile creation Database Configuration, and select the Use a file store for Messaging

50

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch03.fm

Engine (MEs) checkbox. When this profile is used, file stores are used for BPE and SCA service integration buses.

Set data buffer sizes (discardable or cached)

The DiscardableDataBufferSize is the size in bytes of the data buffer used when processing best-effort non-persistent messages. The purpose of the discardable data buffer is to hold message data in memory, because this data is never written to the data store for this Quality of Service. Messages that are too large to fit into this buffer are discarded. The CachedDataBufferSize is the size in bytes of the data buffer used when processing all messages other than best-effort non-persistent messages. The purpose of the cached data buffer is to optimize performance by caching in memory data that might otherwise need to be read from the data store. The DiscardableDataBufferSize and CachedDataBufferSize can be set in the administrative console by selecting Service Integration-Buses bus_name Messaging Engines messaging_engine_name Additional properties Custom properties.

Use a high performance DBMS for messaging engine data stores

For better performance, the messaging engine data stores should use production quality databases, such as DB2. The choice can be made at profile creation time using the advanced profile creation option.

Setting up the data stores for the messaging engines

For information about setting up the data stores see the Information Center article Messaging engines and data stores at the following website: http://publib.boulder.ibm.com/infocenter/wasinfo/v7r0/index.jsp?topic=/com.ibm.web sphere.pmc.nd.multiplatform.doc/tasks/tjm0005_.html

Create the DB2 database and load the data store schemas
Instead of having a DB2 database per messaging engine we put all messaging engines into the same database using different schemas to separate them, as shown in Table 3-1.
Table 3-1 Messaging engine schemas Schema SCASYS SCAAPP CEIMSG BPCMSG Messaging engine box01-server1.SCA.SYSTEM.box01Node01Cell.Bus box01-server1.SCA.APPLICATION. box01Node01Cell.Bus box01-server1.CommonEventInfrastructure_Bus box01-server1.BPC.box01Node01Cell.Bus

Follow these steps: 1. Create one schema definition for each messaging engine with the following command: WAS Install\bin\sibDDLGenerator.bat -system db2 -version 8.1 -platform windows -statementend ; -schema BPCMSG -user user >createSIBSchema_BPCMSG.ddl In this example (used on a Windows operating system), WAS Install represents the IBM BPM Installation directory and user represents the user name. 2. Repeat the command for each schema/messaging engine. 3. To distribute the database across several disks, edit the created schema definitions and put each table in a tablespace named after the schema used. For example, SCAAPP

Chapter 3. Performance tuning and configuration

51

4784ch03.fm

Draft Document for Review January 27, 2012 7:40 am

becomes SCANODE_TS, CEIMSG becomes CEIMSG_TS and so on. The schema definition should look like Example 3-1 after editing.
Example 3-1 Schema definition

CREATE SCHEMA CEIMSG; CREATE TABLE CEIMSG.SIBOWNER ( ME_UUID VARCHAR(16), INC_UUID VARCHAR(16), VERSION INTEGER, MIGRATION_VERSION INTEGER ) IN CEIMSG_TB; CREATE TABLE CEIMSG.SIBCLASSMAP ( CLASSID INTEGER NOT NULL, URI VARCHAR(2048) NOT NULL, PRIMARY KEY(CLASSID) ) IN CEIMSG_TB; . It is possible to provide separate tablespaces for the various tables here. Optimal distribution depends on application structure and load characteristics. In this example one tablespace per data store was used. 4. After creating all schema definitions and defined tablespaces for the tables, create a database named SIB. 5. Create the tablespaces and distribute the containers across available disks by issuing the following command for a system managed tablespace: DB2 CREATE TABLESPACE CEIMSG_TB MANAGED BY SYSTEM USING( '<path>\CEIMSG_TB' ) Place the database log on a separate disk if possible. 6. Create the schema of the database by loading the four schema definitions into the database. For more information about database and DB2-specific tuning, see the following sections: 3.10, Database: General tuning on page 67 3.11, Database: DB2-specific tuning on page 69

Create the data sources for the messaging engines

Create a data source for each messaging engine and configure each messaging engine to use the new data store using the administrative console. Follow these steps: 1. Create a new JDBC provider of type DB2 Universal JDBC Driver Provider for the non-XA data sources if it does not already exist. The XA DB2 JDBC Driver Provider should exist if BPC was configured correctly for DB2. 2. Create four new JDBC data sources, one for CEI as an XA data source, the remaining three as single-phase commit (non-XA) data sources. Table 3-2 provides new names.
Table 3-2 New data sources Name of data source CEIMSG_sib JNDI Name jdbc/sib/CEIMSG Type of JDBC provider DB2 Universal (XA)

52

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch03.fm

SCAAPP_sib SCASYSTEM_sib BPCMSG_sib

jdbc/sib/SCAAPPLICATION jdbc/sib/SCASYSTEM jdbc/sib/BPCMSG

DB2 Universal DB2 Universal DB2 Universal

When creating a data source, perform the following tasks. 1. Clear the Use this Data Source in container managed persistence (CMP) checkbox. 2. Set a component-managed authentication alias. 3. Set the database name to the name used for the database created earlier for messaging, (for example, SIB). 4. Select a driver type of type 2 or type 4. Per DB2 recommendations, use the JDBC Universal Driver type 2 connectivity to access local databases and type 4 connectivity to access remote databases. A type 4 driver requires a host name and valid port to be configured for the database.

Change the data stores of the messaging engines

Use the administrative console to change the data stores of the messaging engines. 1. In the Navigation Panel go to Service Integration Buses and change the data stores for each bus/messaging engine displayed. 2. Put in the new JNDI and schema name for each data store. Clear the Create Tables checkbox, because the tables have already been created. The server immediately restarts the messaging engine. The SystemOut.log shows the results of the change and shows if the messaging engine starts successfully. 3. Restart the server and validate that all systems come up using the updated configuration. The last remaining task is the tuning of the database. For further information about database and DB2-specific tuning, see the following sections: 3.10, Database: General tuning on page 67 3.11, Database: DB2-specific tuning on page 69

3.4.5 Clustered topology tuning

One reason for deploying a clustered topology is to add more resources to system components that are bottlenecked due to increasing load. Ideally, it should be possible to scale up a topology arbitrarily to match the required load. The IBM BPM Network Deployment (ND) infrastructure provides this capability. However, effective scaling still requires standard performance monitoring and bottleneck analysis techniques. The following list details some considerations and tuning guidelines when expanding or tuning a clustered topology. In the discussion, we assume additional cluster members also imply additional server hardware. If deploying more than one cluster member (JVM) on a single physical system, it is important to monitor not just the resource utilization (core, disk, network, and so forth) of the system as a whole, but also the utilization by each cluster member. This allows the detection of a system bottleneck due to a particular cluster member. If all members of a cluster are bottlenecked, scaling can be achieved by adding one or more members to the cluster, backed by appropriate physical hardware. If a singleton server or cluster member is the bottleneck, there are some additional considerations:

Chapter 3. Performance tuning and configuration

53

4784ch03.fm

Draft Document for Review January 27, 2012 7:40 am

A messaging engine in a cluster with a One of N policy (to preserve event ordering) might become the bottleneck. Scaling options include: Hosting the active cluster member on a more powerful hardware server, or removing extraneous load from the existing server. If the messaging engine cluster is servicing multiple busses, and messaging traffic is spread across these busses, consider employing a separate messaging engine cluster per bus. If a particular bus is a bottleneck, consider whether destinations on that bus can tolerate out of order events. In this case, the cluster policy can be changed to allow workload balancing with partitioned destinations. Partitioning a bus also has considerations for balancing work across the messaging engine cluster members. For further information, see the following website: http://publib.boulder.ibm.com/infocenter/wasinfo/v7r0/index.jsp?topic=/co m.ibm.websphere.pmc.nd.doc/concepts/cjt0014_.html A database server might become the bottleneck. Consider the following approaches: If the database server is hosting multiple databases that are active (for example, BPEDB, twproc, twperfdb, and MEDB), consider hosting each database on a separate server. If a single database is driving load, consider a more powerful database server. Beyond these items, database partitioning and clustering capabilities can be exploited.

3.4.6 Web services tuning

If the target of the Web Services import binding is hosted locally in the same application server, the performance can be further improved by exploiting the optimized communication path provided by the Web container. Normally, requests from the Web Services clients are sent through the network connection between the client and the service provider. For local Web services calls, however, WebSphere Application Server offers a direct communication channel bypassing the network layer completely. Perform the following steps to enable this optimization. Use the administrative console to make these changes. 1. Set the Web container custom property enableInProcessConnections to true at Application servers server_name Container Settings Web Container Settings Web container Additional Properties Custom Properties. Do not use wildcard (*) for the host name of the Web Container port. Replace it with the hostname or IP address. The property can be accessed from Application servers messaging engine Container Settings Web Container Settings Web container Additional Properties Web container transport chains WCInboundDefault TCP inbound channel (TCP_2) Related Items Ports WC_defaulthost Host. 2. Use the localhost instead of host name in the Web Services client binding. If the actual host name is used (even if it is aliased to localhost), this optimization is disabled. The property can be accessed from Enterprise Applications application_name Manage Modules application EJB jar Web services client bindings Preferred port mappings binding_name. Use the localhost (for example, localhost:9080) in the URL. 3. Make sure there is not an entry for your server hostame and IP address in your servers hosts file for name resolution. An entry in the hosts file inhibits this optimization by adding name resolution overhead.

54

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch03.fm

3.4.7 Power management tuning

Power management is becoming common in processor technology. Both Intel and Power core processors now have this capability. This capability delivers obvious benefits, but it can also decrease system performance when a system is under high load, so consider whether or not to enable power management. Using POWER6 hardware as an example, ensure that Power Saver Mode is not enabled unless desired. One way to modify or check this setting on AIX is through the Power Management window on the HMC.

3.4.8 Set AIX threading parameters

The IBM JVM threading and synchronization components are based upon the AIX POSIX-compliant Pthreads implementation. The following environment variable settings have been found to improve Java performance in many situations. The variables control the mapping of Java threads to AIX Native threads, turn off mapping information, and allow for spinning on mutex (mutually exclusive) locks: export AIXTHREAD_COND_DEBUG=OFF export AIXTHREAD_MUTEX_DEBUG=OFF export AIXTHREAD_RWLOCK_DEBUG=OFF export AIXTHREAD_SCOPE=S export SPINLOOPTIME=2000

3.5 Business Process Choreographer tuning (for BPEL Business Processes)

This section provides advanced tuning tips for Business Process Choreographer.

3.5.1 Tuning Work-Manager-based navigation for business processes

Work-manager-based navigation is the default navigation mode for BPEL business processes (versus JMS-based navigation). Work-manager-based navigation provides two performance optimizations, keeping the quality of service of process navigation consistent with persistent messaging (JMS-based navigation): Work-Manager-based navigation. A WorkManager is a thread pool of J2EE threads. WorkManager process navigation exploits an underlying capability of WebSphere Application Server to start the processing of ready-to-navigate business flow activities without using messaging as provided by JMS providers. The InterTransactionCache, a part of the Work-Manager-based navigation mode that holds process instance state information in memory, reducing the need to retrieve information from the BPE database. There are several parameters that control usage of these two optimizations. The first set of these parameters are found in the administrative console by navigating to Application Servers server_name Business Integration Business Process Choreographer Business Flow Manager Business Process Navigation Performance. Note that enabling this capability at the cluster level overrides the settings for a specific server. In other words, in a clustered environment it is easiest to enable this capability at the cluster lever.

Chapter 3. Performance tuning and configuration

55

4784ch03.fm

Draft Document for Review January 27, 2012 7:40 am

The key parameters are as follows: Enable advanced performance optimization Select this property to enable both the Work-Manager-based navigation and InterTransactionCache optimizations. Work-Manager-Based Navigation Message Pool Size This property specifies the size of the cache used for navigation messages that cannot be processed immediately, provided Work-Manager-based navigation has been enabled. The cache defaults to a size of (10 * thread pool size of the BPENavigationWorkManager) messages. If this cache reaches its limit, JMS-based navigation will be used for new messages. For optimal performance ensure this Message Pools size is set to a sufficiently high value. InterTransaction Cache Size This property specifies the size of the cache used to store process state information that has also been written to the BPE database. It should be set to twice the number of parallel running process instances. The default value for this property is the thread pool size of the BPENavigationWorkManager. In addition, you can customize the number of threads for the work manager using these settings by selecting Resources Asynchronous Beans Work Managers BPENavigationWorkManager. The minimum and maximum number of threads should be increased from their default values of 5 and 12, respectively, using the methodology outlined in 3.4.3, Tuning for maximum concurrency on page 48. If the thread pool size is modified, the work request queue size should also be modified and set to be twice the maximum number of threads.

3.5.2 Tuning the business process container for JMS navigation

If JMS-based navigation is configured, the following resources need to be optimized for efficient navigation of business processes: Activation specification BPEInternalActivationSpec The Maximum Concurrent Endpoints parameter specifies the parallelism that is used for process navigation across all process instances. Increase the value of this parameter to increase the number of business processes executed concurrently. This resource can be found in the administrative console by selecting Resources Activation Specifications BPEInternalActivationSpec. JMS connection factory BPECFC Set the connection pool size to the number of threads in the BPEInternalActivationSpec plus 10%. This resource can be found in the administrative console at Resources JMS Connection factories BPECFC Connection pool properties. This connection factory is also used when work-manager based navigation is in use, but only for error situations or if the server is highly overloaded.

3.5.3 Tuning task list and process list queries

Task list and process list queries in Business Process Choreographer applications are made using the standard query API (query() and queryAll() APIs, and related REST and Web services interfaces), and the query table API (queryEntities() and queryRows() APIs). All task list and process list queries result in SQL queries against the Business Process

56

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch03.fm

Choreographer database. These SQL queries might need special tuning to provide optimal response times: Up-to-date database statistics are key for good SQL query response times. Databases offer tools to tune SQL queries. In most cases, additional indexes improve query performance with some potential impact on process navigation performance. For DB2, the DB2 design advisor can be used to guide in choosing indexes.

3.5.4 Tuning Business Process Choreographer API calls

Business Process Choreographer API calls are triggered by requests external to the IBM BPM server runtime. Examples are remote EJB requests, Web service requests, Web requests over HTTP, requests that come through the SCA layer, or JMS requests. The connection pools associated with each of these communication mechanisms might need to be tuned. Consider the following hints when tuning the connection pools: API calls for task list and process list queries might take more time to respond, depending on the tuning of the database and the amount of data in the database. Ensure that concurrency (parallelism) is sufficiently high to handle the load and to utilize the CPU. However, increasing the parallelism of API call execution beyond what is necessary can negatively influence response times. Also, increased parallelism can put excessive load on the BPC database. When tuning the parallelism of API calls, measure response times before and after tuning, and adjust the parallelism if necessary.

3.5.5 Tune intermediate components for concurrency

If the input business object is handled by a single thread from end to end, the tuning for the edge components is normally adequate. In many situations, however, there are multiple thread switches during the end-to-end execution path. It is important to tune the system to ensure adequate concurrency for each asynchronous segment of the execution path. Asynchronous invocations of an SCA component utilize an MDB to listen for incoming events that arrive in the associated input queue. Each SCA module defines an MDB and its corresponding activation specification (JNDI name: sca/module name/ActivationSpec). The SCA module MDB is shared by all asynchronous SCA components within the module, including SCA export components. Take this into account when configuring the ActivationSpecs maxConcurrency property value. SCA module MDBs use the same default thread pool as those for JMS exports. The asynchrony in a long-running business process occurs at transaction boundaries (see 2.3.6, Transactionality considerations on page 30 for more details on settings that affect transaction boundaries). BPE defines an internal MDB and its ActivationSpec: BPEInternalActivationSpec. The maxConcurrency parameter needs to be tuned following the same guideline as for SCA module and JMS export MDBs (described in 3.4.3, Tuning for maximum concurrency on page 48). The only catch is there is one BPEInternalActivationSpec for a single IBM BPM server.

3.6 BPMN Business Process Tuning

In addition to the Common tuning guidance above, this section provides tuning guidance specific to BPMN business processes. Note that much of this tuning applies to the Performance Server and SIB databases.

Chapter 3. Performance tuning and configuration

57

4784ch03.fm

Draft Document for Review January 27, 2012 7:40 am

3.6.1 Database tuning

Increase log file size for twproc database (to 16384 pages) Enable filesystem caching for twproc database: db2 alter tablespace userspace1 file system caching Exclude the table, SIBOWNER, from automatic runstats execution as described in: http://www-01.ibm.com/support/docview.wss?uid=swg21452323 Ensure that database statistics are up to date. If running at high throughput rates, consider disabling some or all database auto maintenance tasks to avoid impacting peak throughput. However, if you disable these capabilities, be sure to perform runstats regularly to update database statistics Create new indexes as described in the technote at: http://www-01.ibm.com/support/docview.wss?uid=swg21474536 Create the indexes using the following commands (if using DB2; other databases have similar commands): db2 "CREATE INDEX IDX_SOAB_BPD_INSTANCE ON LSW_BPD_INSTANCE (SNAPSHOT_ID ASC, BPD_INSTANCE_ID ASC) ALLOW REVERSE SCANS COLLECT SAMPLED DETAILED STATISTICS db2 "CREATE INDEX IDX_SOAB_BPD_INSTANCE_VAR ON LSW_BPD_INSTANCE_VARIABLES (BPD_INSTANCE_ID ASC, VARIABLE_TYPE ASC, ALIAS ASC) ALLOW REVERSE SCANS COLLECT SAMPLED DETAILED STATISTICS"

3.6.2 BPD Queue Size and Worker Thread Pool

When running at high throughput rates, the BPD Queue Size and Worker Thread Pool may need to be set to values larger than their defaults (20 and 40, respectively). These values are defined in the file 80EventManager.xml in the configuration directory for the process server. The specific configuration directory is: %BPM%/profiles/<profileName>/config/cells/<cellName>/nodes/<nodeName>/servers/serv er1/process-center or process-server/config/system To change the values, directly edit that file. Here are a couple of guidelines for tuning these parameters: As a rule of thumb, start with a BPD Queue Size (bpd-queue-capacity) of 10 per physical processor core (for example, 40 for a 4 processor core configuration). Tune as needed after that, based on your systems performance. Ensure the Worker Thread Pool size (max-thread-pool-size) is always greather than the sum of all four queue sizes (BPD, Sync, Async, System)

58

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch03.fm

3.7 WebSphere ESB tuning

Following are additional configuration options that are relevant to tuning WebSphere ESB. See 4.2, WebSphere ESB settings on page 87 for a suggested set of initial values to use.

3.7.1 Tune the database if using persistent messaging

If you are using persistent messaging, the configuration of your database becomes important. Use a remote DB2 instance with a fast disk array as the database server. You might find benefit in tuning the connection pooling and statement cache of the data source. For further information about tuning DB2, see the following sections: 3.10, Database: General tuning on page 67 3.11, Database: DB2-specific tuning on page 69 Note the relevant references in Related publications on page 89.

3.7.2 Disable event distribution for CEI

The Event Server that manages events can be configured to distribute events and log them to the event database. Some mediations only require events to be logged to a database. For these cases, performance is improved by disabling event distribution. Because the event server might be used by other applications it is important to check that none of them use event monitoring, which requires event distribution before disabling this. Event distribution can be disabled from the administrative console by selecting Service integration Common Event Infrastructure Event service Event services Default Common Event Infrastructure event server. Clear the Enable event distribution checkbox.

3.7.3 Configure WSRR cache timeout

WebSphere Service Registry and Repository is used by WebSphere ESB for endpoint lookup. When accessing the WebSphere Service Registry and Repository (for example, using the endpoint lookup mediation primitive), results from the registry are cached in WebSphere ESB. The lifetime of the cached entries can be configured from the administrative console by selecting Service Integration WSRR Definitions WSRR_definition name Timeout of Cache. Validate that the timeout is a sufficiently large value. The default timeout is 300 seconds, which is reasonable from a performance perspective. Too low a value results in frequent lookups to the WebSphere Service Registry and Repository, which can be expensive (especially if retrieving a list of results), and includes the associated network latency if the registry is located on a remote machine.

3.8 IBM Business Monitor tuning

This section provides advanced tuning recommendations for IBM Business Monitor.

Chapter 3. Performance tuning and configuration

59

4784ch03.fm

Draft Document for Review January 27, 2012 7:40 am

3.8.1 Configure Java heap sizes

The default maximum heap size in most implementations of Java is too small for many of the servers in this configuration. The Monitor Launchpad installs Monitor and its prerequisite servers with larger heap sizes, but you might check that these sizes are appropriate for your hardware and workload.

3.8.2 Configure CEI

By default, when an event arrives at CEI, it is delivered to the registered consumer (in this case, a particular Monitor Model) and into an additional default queue. Performance is improved by avoiding this double-store, which can be in the administrative console by removing the All Events event group found by going to Service Integration Common Event Infrastructure Event Service Event Services Default Common Event Infrastructure event server Event Groups. Beyond its persistent delivery of events to registered consumers, CEI offers the ability to explicitly store events in a database. This has significant performance overhead and should be avoided if this additional functionality is not needed. The CEI Data Store is also configured in the administrative console by going to Service Integration Common Event Infrastructure Event Service Event Services Default Common Event Infrastructure event server. Clear the Enable Data Store checkbox.

3.8.3 Configure message consumption batch size

Consuming events in large batches is much more efficient than one at a time. Up to some limit, the larger the batch size, the higher is the event processing throughput. But there is a trade-off: consuming events, processing them, and persisting them to the Monitor database is done as a transaction. So, while a larger batch size yields better throughput, it costs more if you have to roll back. If you experience frequent rollbacks, consider reducing the batch size. This can be done in the administrative console under the server scope. Select Applications Monitor Models version Runtime Configuration Tuning Message Consumption Batch size: <default 100>.

3.8.4 Enable KPI caching

The cost of calculating aggregate KPI values increases as completed process instances accumulate in the database. A KPI cache is available to reduce the overhead of these calculations, at the cost of some staleness in the results. The refresh interval is configurable in the WebSphere administrative console. Select Applications Monitor Models version Runtime Configuration KPI KPI Cache Refresh Interval. A value of zero (the default) disables the cache.

3.8.5 Use table-based event delivery

There are two ways in which events can be delivered by CEI to a monitor model via a JMS queue, or via a database table. This is chosen at application install time for a monitor model. It is strongly recommended that you choose table-based event delivery (sometimes called queue bypass), both for reliability and for performance/scalability reasons.

60

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch03.fm

3.8.6 Enable the Data Movement Service

By default, the same tables are used for event processing and for dashboard reporting. There is an optional, scheduled service, called the Data Movement Service (DMS) that can be enabled to switch dashboard reporting to be against a separate set of tables, and to periodically copy the data from the servers event processing tables to the dashboard reporting tables. In this mode, the indexes on each are optimized; the servers event processing tables are optimized for quick insert/update/delete, and the dashboard reporting tables are optimized for common dashboard-related queries. It is strongly recommended that DMS be enabled in any production scenario

3.9 Business Space Tuning

Since Business Space (BSpace) is an AJAX application, much of its code is executed in the browser via Javascript and its performance is therefore highly dependent on the browser and the client system it is running on. In order to improve response times in Business Space, you might consequently need to adjust server- and client settings. This chapter covers the key tuning parameters for BSpace solutions.

3.9.1 Ensure Browser Cache is Enabled

By default, Business Space allows the browser to cache static information like Stylesheets and Javascript files for 24 hours. However, if the browser is configured in a way that it does not cache data, the same resources will be loaded over and over again resulting in long page load times. To enable caching, follow these steps for your browser: Internet Explorer Go to: Tools Internet Options General Browsing History Set the cache size (Disk space to use) to at least 50 MB. Firefox Follow this path in the menu: Tools Options Advanced Network Offline Storage Set the offline storage to at least 50 MB.

Ensure Browser Cache is not Cleared on Logout

Modern browsers have options to clear the cache on exit; make sure this setting isn't enabled. Instructions for IE8 and FF follow: Use the following procedure for IE8. 1. Click on Tools Internet Options. 2. On the General tab, make sure the "Delete browser history on exit" checkbox is unchecked, as shown in Figure 3-1 on page 62.

Chapter 3. Performance tuning and configuration

61

4784ch03.fm

Draft Document for Review January 27, 2012 7:40 am

Figure 3-1 Delete browsing history

3. Click on the Advanced tab and make sure the "Empty Temporary Internet Files folder when browser is closed" checkbox is unchecked, as shown in Figure 3-2 on page 63.

62

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch03.fm

Figure 3-2 Empty Temporary Internet Files folder

4. Click OK to save the settings. You should restart your browser to make sure the changes have taken effect. Use the following procedure for Firefox 3.6. 1. Select Tools Options. 2. Select the Privacy tab and make sure it says Firefox will: Remember History, as shown in Figure 3-3 on page 64.

Chapter 3. Performance tuning and configuration

63

4784ch03.fm

Draft Document for Review January 27, 2012 7:40 am

Figure 3-3 Remember history

3. If Remember history is the setting, then you are done (caching is enabled properly). If "Never remember history" is selected, caching is disabled and the setting needs to be changed. If "Use custom settings for history" is selected, there are a couple options that can be selected that will still allow caching, as shown in Figure 3-4 on page 64.

Figure 3-4 Use custom settings for history

64

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch03.fm

a. Make sure the private browsing session is disabled (screenshot above shows it enabled, this checkbox should be unchecked). This means caching is enabled, but once the browser is closed everything gets erased (not just cache but browsing history, etc it is as if you were never using the browser). b. If private browsing is not selected, then make sure the "Clear history when Firefox closes" box is unchecked, as shown in Figure 3-5 on page 65.

Figure 3-5 Clear history

4. Click OK to save your settings and restart the browser to make sure changes have taken effect.

3.9.2 Optimize Complex Task Message Performance

When opening tasks with large and complex input messages via the Task Information widget, your browser might take a long time to render and display the task. If you are facing this issue, refer to the following technote and apply the practices described there: http://www-01.ibm.com/support/docview.wss?uid=swg21468931&wv=1

3.9.3 Tune the HTTP Server

Business Space in production scenarios is generally deployed using a topology which includes a HTTP server or an HTTP server plugin. As an AJAX client, Business Space will naturally send multiple requests to the server, which are then handled by this HTTP server. In order to ensure good end-user response times, tune your HTTP server to efficiently handle the expected load. For further information on tuning the IBM HTTP Server, which is included as part of BPM 7.5.0, see the following:

Chapter 3. Performance tuning and configuration

65

4784ch03.fm

Draft Document for Review January 27, 2012 7:40 am

http://publib.boulder.ibm.com/httpserv/ihsdiag/ihs_performance.html For further information on tuning BSpace, also see the Business Space performance and scalability paper: http://www-01.ibm.com/support/docview.wss?uid=swg27020684&wv=1

3.9.4 Optimize Performance when not using Federation Mode

Those familiar with the performance characteristics of Human Workflow widgets with WebSphere Process Server (WPS) v7 may experience an out-of-the-box performance degradation when using Human Workflow widgets such as Tasks, Processes and Task Definitions in BPM 7.5.0. This degradation should only be observed if using Business Space on a new (not a migrated) installation of BPM 7.5.0. These performance considerations apply if you use the Human Task Management widgets with only Business Process Definition (BPD) processes and human services on a single application server, or on a cluster. This degradation is caused by the addition of a federation service layer and the higher number of items to be retrieved, which is a usability enhancement in BPM 7.5 that enables displaying all tasks in a single unified task list. Note: This does not apply to an upgrade or migration from WPS v7. To check the Business Space target services configuration in the administration console, complete the following steps. 1. Open the IBM Business Process Manager Integrated Solutions Console. 2. Select the Servers tab in the navigation bar. 3. Select WebSphere. 4. Select Application Servers server type and choose the application server on which you want to run Business Space. The page to configure an application server opens. 5. In the Business Integration section on the Configuration tab select the Business Space Configuration entry. The Business Space Configuration page opens. 6. In the Additional Properties section, select the REST service endpoint registration entry. The REST service endpoint registration page opens. 7. Check the Service Endpoint Target values for the Process services and Task services REST endpoint types. The default out-of-the-box setting is Federated REST services. To resolve the problem, complete the following steps. 1. Change the Business Space target services configuration in the administration console by following the steps in the Diagnosing the problem section. 2. Change the Process services to the Business Process Choreographer REST services endpoint of the application server that is running your processes and human tasks. 3. Change the Tasks services to the Business Process Choreographer REST services endpoint of the application server that is running your processes and human tasks. 4. Click OK to save the configuration changes. 5. Log out of Business Space and then log in again.

66

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch03.fm

Note: To use only the BPD processes and human services of a single application server or cluster, change the Process services and Task services to the Business process definition engine REST services of the application server that is running your BPD processes and human services.

3.10 Database: General tuning

This section provides general tuning hints for databases.

3.10.1 Provide adequate statistics for optimization

Databases often have a wide variety of available choices when determining the best approach to accessing data. Statistics, which describe the shape of the data, are used to guide the selection of a low-cost data access strategy. Statistics are maintained on tables and indexes. Examples of statistics include the number of rows in a table and the number of distinct values in a certain column. Gathering statistics can be expensive, but fortunately for many workloads, a set of representative statistics allows for good performance over a large span of time. It might be necessary to refresh statistics periodically if the data population shifts dramatically.

3.10.2 Place database log files on a fast disk subsystem

Databases are designed for high availability, transactional processing and recoverability. Because for performance reasons changes to table data might not be written immediately to disk. These changes are made recoverable by writing to the database log. Updates are made to database log files when the log buffer fills, at transaction commit time, and for some implementations after a maximum interval of time. As a result, database log files might be heavily utilized. More importantly, log writes hold commit operations pending, meaning that the application is synchronously waiting for the write to complete. Therefore, write access performance to the database log files is critical to overall system performance. We recommend that database log files be placed on a fast disk subsystem with write back cache.

3.10.3 Place logs on a separate device from the tablespace containers

A basic strategy for all database storage configurations is to place the database logs on dedicated physical disks, ideally on a dedicated disk adapter. This reduces disk access contention between I/O to the tablespace containers and I/O to the database logs and preserves the mostly sequential access pattern of the log stream. Such separation also improves recoverability when log archival is employed.

3.10.4 Provide sufficient physical memory

Accessing data in memory is much faster than reading it from disk. With 64-bit hardware being readily available and memory prices continuing to fall, it makes sense to provision enough memory to avoid most disk reads in steady state for many performance critical workloads.

Chapter 3. Performance tuning and configuration

67

4784ch03.fm

Draft Document for Review January 27, 2012 7:40 am

Great care should be taken to avoid virtual memory paging in the database machine. The database manages its memory with the assumption that it is never paged, and does not cooperate well should the operating system decide to swap some of its pages to disk.

3.10.5 Avoid double buffering

Because the database attempts to keep frequently accessed data in memory, in most cases there is no benefit to using file system caching. In fact, performance typically improves by using direct I/O, when files read by the database bypass the file system cache and only one copy of the data is held in memory. This allows more memory to be given to the database and avoids overheads in the file system as it manages its cache. A further advantage can be gained on some operating systems such as AIX by using concurrent I/O. This bypasses per-file locking, shifting responsibility for concurrency control to the database and in some cases allowing more useful work to be offered to the adapter or the device. An important exception to this guideline occurs for large objects (LOB, BLOB, CLOB, and so forth) that are not buffered by the database itself. In this case it can be advantageous to arrange for file system caching, preferably only for files that back large objects.

3.10.6 Refine table indexes as required

IBM BPM products typically provide a reasonable set of indexes for the database tables they use. In general, creating indexes involves a tradeoff between the cost of queries and the cost of statements which insert, update, or delete data. For query intensive workloads, it makes sense to provide a rich variety of indexes as required to allow rapid access to data. For update intensive workloads, it is often helpful to minimize the number of indexes defined, as each row modification might require changes to multiple indexes. Note that indexes are kept current even when they are infrequently used. Index design therefore involves compromises. The default set of indexes might not be optimal for the database traffic generated by a BPM product in a specific situation. If database CPU or disk utilization is high or there are concerns with database response time, it might be helpful to consider changes to indexes. DB2 and Oracle databases provide assistance in this area by analyzing indexes in the context of a given workload. Recommendations are given to add, modify, or remove indexes. One caveat is that if the workload does not capture all relevant database activity, a necessary index might appear unused, leading to a recommendation that it be dropped. If the index is not present, future database activity could suffer as a result.

3.10.7 Archive completed process instances

Over time, completed process instances accumulate in the servers databases. This accumulation can alter the performance characteristics of the solution being measured. It is helpful to archive completed process instances in order to ensure that database population is controlled. Archiving procedures can be found in the technote available at the following website: http://www-01.ibm.com/support/docview.wss?uid=swg21439859

68

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch03.fm

3.11 Database: DB2-specific tuning

Providing a comprehensive DB2 tuning guide is beyond the scope of this paper. However, there are a few general rules of thumb that can assist in improving the performance of DB2 environments. In the following sections, we discuss these rules, and provide pointers to more detailed information. The complete set of current DB2 manuals (including database tuning guidelines) can be found in the DB2 Information Center: http://publib.boulder.ibm.com/infocenter/db2luw/v9r7/index.jsp Another excellent reference is the developerWorks article Best practices for DB2 for Linux, UNIX, and Windows, available a the following website: http://www.ibm.com/developerworks/data/bestpractices/

3.11.1 Update database statistics

DB2 provides an Automatic Table Maintenance feature, which runs the RUNSTATS command in the background as required. This ensures that the correct statistics are collected and maintained. This is controlled by the database configuration parameter auto_runstats, and is enabled by default for databases created by DB2 V9.1 and beyond. See the Configure Automatic Maintenance... wizard at the database level in the DB2 Control Center. One approach to updating statistics manually on all tables in the database is to use the REORGCHK command. Dynamic SQL, such as that produced by JDBC, immediately takes the new statistics into account. Static SQL, such as that in stored procedures, must be explicitly rebound in the context of the new statistics. Example 3-2 is an example that performs these steps to gather basic statistics on database DBNAME:
Example 3-2 Gathering basic statistics on database DBNAME example

db2 connect to DBNAME db2 reorgchk update statistics on table all db2 connect reset db2rbind DBNAME all The REORGCHK and rebind should be executed when the system is relatively idle so that a stable sample might be acquired and to avoid possible deadlocks in the catalog tables. It is better to gather additional statistics, so consider also using the following command for every table requiring attention: runstats on table <schema>.<table> with distribution and detailed indexes

3.11.2 Set buffer pool sizes correctly

A buffer pool is an area of memory into which database pages are read, modified, and held during processing. Buffer pools improve database performance. If a needed page of data is already in the buffer pool, that page is accessed faster than if the page had to be read directly from disk. As a result, the size of the DB2 buffer pools is critical to performance. The amount of memory used by a buffer pool depends upon two factors: Size of buffer pool pages Number of pages allocated

Chapter 3. Performance tuning and configuration

69

4784ch03.fm

Draft Document for Review January 27, 2012 7:40 am

Buffer pool page size is fixed at creation time and might be set to 4, 8, 16 or 32 KB. The most commonly used buffer pool is IBMDEFAULTBP which has a 4 KB page size. All buffer pools reside in database global memory, allocated on the database machine. The buffer pools must coexist with other data structures and applications, all without exhausting available memory. In general, having larger buffer pools improves performance up to a point by reducing I/O activity. Beyond that point, allocating additional memory no longer improves performance. DB2 V9.1 and later provide self tuning memory management, which includes managing buffer pool sizes. This is controlled globally by the self_tuning_mem database level parameter, which is on by default. Individual buffer pools can be enabled for self tuning using SIZE AUTOMATIC at CREATE or ALTER time. To choose appropriate buffer pool size settings manually, monitor database container I/O activity by using system tools or by using DB2 buffer pool snapshots. Be careful to avoid configuring large buffer pool size settings that lead to paging activity on the system.

3.11.3 Maintain proper table indexing

The DB2 Design Advisor, available from the Control Center, provides recommendations for schema changes, including changes to indexes. It can be launched from the menu presented when right-clicking a database in the left column.

3.11.4 Size log files appropriately

When using circular logging, it is important that the available log space permits dirty pages in the bufferpool to be cleaned at a reasonably low rate. Changes to the database are immediately written to the log, but a well tuned database will coalesce multiple changes to a page before eventually writing that modified page back to disk. Naturally, changes recorded only in the log cannot be overwritten by circular logging. DB2 detects this condition and forces the immediate cleaning of dirty pages required to allow switching to a new log file. While this mechanism protects the changes recorded in the log, all application logging must be suspended until the necessary pages are cleaned. DB2 works to avoid pauses when switching log files by proactively triggering page cleaning under control of the database level softmax parameter. The default value of 100 for softmax begins background cleaning activities when the gap between the current head of the log and the oldest log entry recording a change to a dirty page exceeds 100% of one log file in size. In extreme cases this asynchronous page cleaning cannot keep up with log activity, leading to log switch pauses that degrade performance. Increasing the available log space gives asynchronous page cleaning more time to write dirty bufferpool pages and avoid log switch pauses. A longer interval between cleanings allows multiple changes to be coalesced on a page before it is written, which reduces the required write throughput by making page cleaning more efficient. Available logspace is governed by the product of log file size and the number of primary log files, which are configured at the database level. The logfilsiz setting is the number of 4 K pages in each log file. The logprimary setting controls the number of primary log files. The Control Center also provides a Configure Database Logging... wizard. As a starting point, try using 10 primary log files that are large enough that they do not wrap for at least a minute in normal operation.

70

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch03.fm

Increasing the primary log file size has implications for database recovery. Assuming a constant value for softmax, larger log files mean that recovery might take more time. The softmax parameter can be lowered to counter this, but keep in mind that more aggressive page cleaning might also be less efficient. Increasing the softmax parameter gives additional opportunities for write coalescing at the cost of longer recovery time. The default value softmax is 100, meaning that the database manager attempts to clean pages such that a single log file needs to be processed during recovery. For best performance, we recommend increasing this to 300, meaning that 3 log files might need processing during recovery: db2 update db config for yourDatabaseName using softmax 300

3.11.5 Inline LOBs to improve throughput for high volume systems

When tuning database tables, consider setting inline length for LOB columns to potentially improve performance when running at high throughput rates. This parameter enables DB2 to optimize database access by storing the LOBs on the same page as the database row if the LOB will fit within the inline length provided. This can improve the performance of query and retrieval operations. Note that log traffic is increased when this parameter is set. There are several columns where this can apply. Of particular importance, setting inline length for the DATA column in the LSW_BPD_INSTANCE_DATA table. IBMs internal peformance evaluation shows that this can significantly improve throughput for high volume production systems.

3.11.6 Use SMS for tablespaces containing large objects

When creating REGULAR or LARGE tablespaces in DB2 V9.5 and later that contain performance critical LOB data, we recommend specifying MANAGED BY SYSTEM to gain the advantages of cached LOB handling in SMS. Among IBM BPM products, this consideration applies to: The BPEL Process Choreographer database, sometimes called BPEDB The BPMN databases, sometimes called twproc and twperfdb Databases backing service integration bus messaging engine data stores. For background, see 3.10.5, Avoid double buffering on page 68. A detailed explanation follows. DB2 tablespaces can be configured with NO FILE SYSTEM CACHING, which in many cases improves system performance. If a tablespace is MANAGED BY SYSTEM, then it uses System Managed Storage (SMS), which provides desirable special case handling for LOB data with regard to caching. Even if NO FILE SYSTEM CACHING is in effect (by default or as specified) access to LOB data still uses the file system cache. If a tablespace is MANAGED BY DATABASE, it uses Database Managed Storage (DMS), which does not differentiate between LOB and non-LOB data with regard to caching. In particular, NO FILE SYSTEM CACHING means that LOB access reads and writes directly to disk. Unconditionally reading LOBs from disk can cause high disk utilization and poor database performance. Since V9.1, DB2 has by default created databases that use automatic storage (AUTOMATIC STORAGE YES), meaning that the database manages disk space allocations itself from one or more pools of available file system space called storage paths. If automatic storage is

Chapter 3. Performance tuning and configuration

71

4784ch03.fm

Draft Document for Review January 27, 2012 7:40 am

enabled, CREATE TABLESPACE uses it by default (MANAGED BY AUTOMATIC STORAGE). For non-temporary tablespaces, REGULAR and LARGE, automatic storage is implemented using DMS on files. Before DB2 V9.5 the default caching strategy for tablespaces was FILE SYSTEM CACHING. In V9.5, this was changed to NO FILE SYSTEM CACHING for platforms where direct I/O or concurrent I/O is available. Taking defaults on V9.5 we now have a database with AUTOMATIC STORAGE YES, and a tablespace that is MANAGED BY AUTOMATIC STORAGE and in many cases NO FILE SYSTEM CACHING. Such a tablespace, which is implemented using DMS, does not cache LOBs in the buffer pool or the file system.

3.11.7 Ensure that sufficient locking resources are available

Locks are allocated from a common pool controlled by the locklist database level parameter, which is the number of 4K pages set aside for this use. A second database level parameter, maxlocks, bounds the percentage of the lock pool held by a single application. When an application attempts to allocate a lock that exceeds the fraction allowed by maxlocks, or when the free lock pool is exhausted, DB2 performs lock escalation to replenish the supply of available locks. Lock escalation involves replacing many row locks with a single table-level lock. While lock escalation addresses the immediate problem of lock pool overuse or starvation, it can lead to database deadlocks, and so should not occur frequently during normal operation. In some cases, application behavior can be altered to reduce pressure on the lock pool by breaking up large transactions that lock many rows into smaller transactions. It is usually simpler to try tuning the database first. Beginning with Version 9, DB2 adjusts the locklist and maxlocks parameters automatically by default. To tune these manually, observe whether lock escalations are occurring either by examining db2diag.log or by using the system monitor to gather snapshots at the database level. If the initial symptom is database deadlocks, consider whether these are initiated by lock escalations. Check the Lock escalations count in the output from: db2 get snapshot for database yourDatabaseName Current values for locklist and maxlocks can be obtained by examining the output from: db2 get db config for yourDatabaseName These values can be altered, for example to 100 and 20, like this: db2 update db config for yourDatabaseName using locklist 100 maxlocks 20 When increasing the locklist size, consider the impacts of the additional memory allocation required. Often the locklist is relatively small compared with memory dedicated to buffer pools, but the total memory required must not lead to virtual memory paging. When increasing the maxlocks fraction, consider whether a larger value allows a few applications to drain the free lock pool, leading to a new cause of escalations as other applications needing relatively few locks encounter a depleted free lock pool. Often it is better to start by increasing locklist size alone.

3.11.8 Bound the size of the catalog cache for clustered applications
The catalog cache is used to avoid repeating expensive activities, notably preparing execution plans for dynamic SQL. Thus it is important that the cache be sized appropriately.

72

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch03.fm

By default, several 4 KB pages of memory are allocated for each possible application as defined by the MAXAPPLS database parameter. The multiplier is 4 for DB2 9, and 5 for DB2 9.5 and beyond. MAXAPPLS is AUTOMATIC by default, and its value is adjusted to match the peak number of applications connected at runtime. When running clustered applications, such as those deployed in the BPEL Process Choreographer in IBM BPM 7.5.0, we have observed a value of more than 1000 for MAXAPPLS, meaning that at least 4000 pages would be allocated for the catalog cache given default tuning. For the same workload, 500 pages were sufficient: db2 update db config for yourDatabaseName using catalogcache_sz 500 The default behavior assumes heterogeneous use of database connections. A clustered application typically has more homogeneous use across connections, allowing a smaller package cache to be effective. Bounding the package cache size frees up memory for other more valuable uses. To tune the CATALOGCACHE_SZ database parameter manually, see the recommendations at the following website: http://publib.boulder.ibm.com/infocenter/db2luw/v9/topic/com.ibm.db2.udb.admin.doc /doc/r0000338.htm

3.11.9 (Before DB2 V9.5) Size the database heap appropriately

DB2 Version 9.5 and later provide AUTOMATIC tuning of the database heap by default. We recommend using this when available. To tune the DBHEAP database parameter manually, see the recommendations documented at the following website: http://publib.boulder.ibm.com/infocenter/db2luw/v9/topic/com.ibm.db2.udb.admin.doc /doc/r0000276.htm

3.11.10 (Before DB2 V9.7) Size the log buffer appropriately

Before DB2 Version 9.7 the default LOGBUFSZ was only 8 pages. We recommend setting this to 256, which is the default in Version 9.7: db2 update db config for yourDatabaseName using logbufsz 256

3.11.11 (DB2 V9.7 and later) Consider disabling current commit

DB2 Version 9.7 supports new query semantics that always return the committed value of the data at the time the query is submitted. This support is on by default for newly created databases. We found that performance improved in some cases when we disabled the new behavior, reverting to the original DB2 query semantics: db2 update db config for yourDatabaseName using cur_commit disabled

3.11.12 Recommendations for BPEL business processes in IBM BPM 7.5.0

The following website discusses specifying initial DB2 database settings with examples of creating SMS tablespaces for the BPEDB. It also contains useful links for planning the BPEDB database and fine-tuning the Business Process Choreographer database:

Chapter 3. Performance tuning and configuration

73

4784ch03.fm

Draft Document for Review January 27, 2012 7:40 am

http://publib.boulder.ibm.com/infocenter/dmndhelp/v7r0mx/index.jsp?topic=/com.ibm. websphere.bpc.doc/doc/bpc/t5tuneint_spec_init_db_settings.html The following website discusses creating a DB2 for Linux, UNIX, and Windows database for Business Process Choreographer and gives details on BPEDB database creation, including pointers to useful creation scripts for a production environment: http://publib.boulder.ibm.com/infocenter/dmndhelp/v7r0mx/index.jsp?topic=/com.ibm. websphere.bpc.doc/doc/bpc/t2codbdb.html

3.11.13 Recommendations for Business Monitor 7.5.0

Improve Concurrency by Setting Registry Variables
DB2 allows the deferral of row locks for Cursor Stability (CS) or Read Stability (RS) isolation scans in some cases until a record is known to satisfy the predicates of a query. In most cases, DB2 locks each row before determining if the row meets the query when row locking is performed during a table or index scan. To improve concurrency, it may be possible to defer row locking until after determining that a row qualifies for a query. Usually, concurrency is improved by setting the registry variables DB2_SKIPDELETED and DB2_SKIPINSERTED. Both of these registry variables permit scans to unconditionally skip uncommitted deletes and inserts, respectively. Here is an example which illustrates how to enable these two registry variables for the database MONITOR: db2 connect to MONITOR db2set DB2_SKIPDELETED=ON db2set DB2_SKIPINSERTED=ON

Avoid Transaction Log is Full

DB2 Health Monitor (HMON) captures information about the database manager, database, tablespaces, and tables regularly. It calculates health indicators based on data retrieved from database system monitor elements, the operating system, and the DB2 system. Transactions will hang if HMON examines the status for the table <MONMEBUS_SCHEMA>.SIBOWNER and the transaction occupies the entire transaction log, resulting in an error with SQLCODE 964. To prevent this from happening, you can disable the HMON process using the following command: db2 update dbm cfg using HEALTH_MON OFF

Set Lock Timeout properly

The parameter LOCKTIMEOUT specifies the number of seconds that an application will wait to obtain a lock. This parameter helps avoid global deadlocks for applications. If you set this parameter to 0, locks are not waited for. In this situation, if the lock is unavailable at the time of the request, the application immediately receives a -911 return code. If you set this parameter to -1, lock timeout detection is turned off.

74

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch03.fm

IBMs internal evaluation shows that a value of 30 seconds could be a good starting value; tune as necessary after setting this value. An example on how to set this parameter for the database MONITOR follows: db2 -v update db cfg for MONITOR using LOCKTIMEOUT 30

Limit event XML to 32K where possible

Events that are small enough will be persisted to a regular VARCHAR column in the incoming events table; large events will be persisted to a BLOB column instead. In DB2, the largest VARCHAR column is 32768 bytes. Performance is considerably improved when a VARCHAR column can be used instead of a BLOB.

Use Materialized Query Tables

When you have a very deep history of data (for example, more than 10 million MC instances), the response time for dashboard navigation in dimensional reports can degrade significantly. IBM Business Monitor 7.5 uses Cognos Business Intelligence 10.1 for its dimensional reports. A tool is provided with the product, that can be used with DB2 to generate cube summary tables that pre-compute the values of measures for known dimensional member values (such as the value of the average loan amount measure based on values for the customer loyalty level dimension say for bronze, silve, and gold customers). DB2 calls these materialized query tables, and IBM Business Monitor provides a scheduled service to refresh these on a periodic basis. For details on using this capability, see the Information Center: http://publib.boulder.ibm.com/infocenter/dmndhelp/v7r5mx/index.jsp?topic=/com.ibm. wbpm.mon.admin.doc/data/enable_cubesumtable_refresh.html

3.12 Database: Oracle specific tuning

As with DB2, providing a comprehensive Oracle database tuning guide is beyond the scope of this paper. However, there are a few general rules of thumb that can assist in improving the performance of Oracle environments when used with IBM BPM products. In the following sections, we discuss these rules, and provide pointers to more detailed information. In addition, the following references are useful: Oracle Database 11g Release 1 documentation (includes a Performance Tuning Guide): http://www.oracle.com/pls/db111/homepage A white paper discussing Oracle on AIX: http://www-03.ibm.com/support/techdocs/atsmastr.nsf/WebIndex/WP100883

3.12.1 Update database statistics

Oracle provides an automatic statistics gathering facility, which is enabled by default. One approach to updating statistics manually on all tables in a schema is to use the dbms_stats utility. Consult the Oracle product documentation for further information.

3.12.2 Set buffer cache sizes correctly

Oracle provides automatic memory management for buffer caches. For additional discussion on configuring automatic memory management and for guidance on manually setting buffer cache sizes, see the following references.
Chapter 3. Performance tuning and configuration

75

4784ch03.fm

Draft Document for Review January 27, 2012 7:40 am

For Oracle 10g R2: http://download.oracle.com/docs/cd/B19306_01/server.102/b14211/memory.htm#i29118 For Oracle 11g R1: http://download.oracle.com/docs/cd/B28359_01/server.111/b28274/memory.htm#i29118

3.12.3 Maintain proper table indexing

The SQL Access Advisor, available from the Enterprise Manager, provides recommendations for schema changes, including changes to indexes. It can be found starting at the database home page, then following the Advisor Central link in the Related Links section at the bottom of the page.

3.12.4 Size log files appropriately

Unlike DB2, Oracle performs an expensive checkpoint operation when switching logs. The checkpoint involves writing all dirty pages in the buffer cache to disk. Therefore, it is important to make the log files large enough that switching occurs infrequently. Applications that generate a high volume of log traffic need larger log files to achieve this goal.

3.12.5 Recommendations for BPEL business processes in IBM BPM 7.5.0

The following website discusses specifying initial Oracle database settings: http://publib.boulder.ibm.com/infocenter/dmndhelp/v7r0mx/index.jsp?topic=/com.ibm. websphere.bpc.doc/doc/bpc/t5tuneint_spec_init_db_oracle.html The following website discusses creating an Oracle database for Business Process Choreographer and gives details on BPEDB database creation, including pointers to useful creation scripts for a production environment: http://publib.boulder.ibm.com/infocenter/dmndhelp/v7r0mx/index.jsp?topic=/com.ibm. websphere.bpc.doc/doc/bpc/t2codbdb.html The default Oracle policy for large objects (LOB) is to store the data within the row, when the size of the object does not exceed a threshold. In some cases, workloads have LOBs that regularly exceed this threshold. By default, such LOB accesses bypass the buffer cache, meaning that LOB reads are exposed to disk I/O latencies when using the preferred direct or concurrent path to storage.

3.13 Advanced Java heap tuning

Because the IBM BPM product set is written in Java, the performance of the JVM has a significant impact on the performance delivered by these products. JVMs externalize multiple tuning parameters that might be used to improve both authoring and runtime performance. The most important of these are related to garbage collection and setting the Java heap size. This section deals with these topics in detail. The products covered in this paper utilize IBM JVMs on most platforms (AIX, Linux, Windows, and so forth), and the HotSpot JVMs on selected other systems, such as Solaris and HP/UX. Vendor-specific JVM implementation details and settings are discussed as appropriate. All BPM v7 products in this document use Java 6. It has characteristics similar to Java 5 used in the BPM v 6.1 and v6.2.0 products, but much different from Java 1.4.2 used by V6.0.2.x and 76
IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch03.fm

earlier releases. For brevity, only Java 6 tuning is discussed here. The following website is the IBM Java 6 Diagnostics Guide: http://publib.boulder.ibm.com/infocenter/javasdk/v6r0/index.jsp The guide discusses many more tuning parameters than those discussed in this paper, but most are for specific situations and are not of general use. For a more detailed description of IBM Java 6 garbage collection algorithms, see Memory Management in the chapter titled Understanding the IBM SDK for Java. Sun HotSpot JVM references: The following website provides a useful summary of HotSpot JVM options for Solaris: http://java.sun.com/docs/hotspot/VMOptions.html The following website provides a useful FAQ about the Solaris HotSpot JVM: http://java.sun.com/docs/hotspot/PerformanceFAQ.html#20 For more performance tuning information of Suns HotSpot JVM, consult the following website: http://java.sun.com/docs/performance/

3.13.1 Monitoring garbage collection (GC)

To set the heap correctly, determine how the heap is being used. This is done by collecting a verbosegc trace. A verbosegc trace prints garbage collection actions and statistics to stderr in IBM JVMs and stdout in Sun HotSpot JVMs. The verbosegc trace is activated by using the Java run-time option -verbose:gc. Output from verbosegc is different for the HotSpot and IBM JVMs, as shown by Example 3-3 and Example 3-4.
Example 3-3 Example IBM JVM verbosegc trace output

<af type="tenured" id="12" timestamp="Fri Jan 18 15:46:15 2008" intervalms="86.539"> <minimum requested_bytes="3498704" /> <time exclusiveaccessms="0.103" /> <tenured freebytes="80200400" totalbytes="268435456" percent="29" > <soa freebytes="76787560" totalbytes="255013888" percent="30" /> <loa freebytes="3412840" totalbytes="13421568" percent="25" /> </tenured> <gc type="global" id="12" totalid="12" intervalms="87.124"> <refs_cleared soft="2" threshold="32" weak="0" phantom="0" /> <finalization objectsqueued="0" /> <timesms mark="242.029" sweep="14.348" compact="0.000" total="256.598" /> <tenured freebytes="95436688" totalbytes="268435456" percent="35" > <soa freebytes="87135192" totalbytes="252329472" percent="34" /> <loa freebytes="8301496" totalbytes="16105984" percent="51" /> </tenured> </gc> <tenured freebytes="91937984" totalbytes="268435456" percent="34" > <soa freebytes="87135192" totalbytes="252329472" percent="34" /> <loa freebytes="4802792" totalbytes="16105984" percent="29" /> </tenured> <time totalms="263.195" /> </af>
Example 3-4 Example Solaris HotSpot JVM verbosgc trace output (young and old)

[GC 325816K 83372K(776768K), 0.2454258 secs]

Chapter 3. Performance tuning and configuration

77

4784ch03.fm

Draft Document for Review January 27, 2012 7:40 am

[Full GC 267628K 83769K <- live data (776768K), 1.8479984 secs] Sun HotSpot JVM verbosegc output can be more detailed by setting additional options: -XX:+PrintGCDetails -XX:+PrintGCTimeStamps It is tedious to parse the verbosegc output using a text editor. There are good visualization tools on the Web that can be used for more effective Java heap analysis. The IBM Pattern Modeling and Analysis Tool (PMAT) for Java Garbage Collector is one such tool. It is available for free download at IBM alphaWorks from the following website: http://www.alphaworks.ibm.com/tech/pmat PMAT supports the verbosegc output formats of JVMs offered by major JVM vendors such as IBM, Sun, and HP.

3.13.2 Setting the heap size for most configurations

This section contains guidelines for determining the appropriate Java heap size for most configurations. If your configuration requires that more than one JVM runs concurrently on the same system (for example, if you run both an IBM BPM server and IBM Integration Developer on the same system), you should also read the section, 3.13.3, Setting the Heap Size when running multiple JVMs on one system on page 79. If your objective is to support large business objects, read 3.4.2, Tuning for large objects on page 47. For most production applications, the IBM JVM Java heap size defaults are too small and should be increased. In general the HotSpot JVM default heap and nursery size are also too small and should be increased. There are several approaches to setting optimal heap sizes. We describe the approach that most applications should use when running the IBM JVM on AIX. The essentials can be applied to other systems. Set the initial heap size (-Xms option) to something reasonable (for example, 768 MB on a 64 bit JVM, or 256 MB on a 32 bit JVM), and the maximum heap size (-Xmx) option to something reasonable, but large (for example, 3072 MB on a 64 bit JVM, or 1024 MB on a 32 bit JVM). Of course, the maximum heap size should never force the heap to page. It is imperative that the heap always stays in physical memory. The JVM then tries to keep the GC time within reasonable limits by growing and shrinking the heap. The output from verbosegc is used to monitor GC activity. If Generational Concurrent GC is used (-Xgcpolicy:gencon), the new area size can also be set to specific values. By default, the new size is a quarter of the total heap size or 64 MB, whichever is smaller. For better performance, the nursery size should be 1/2 of the heap size or larger, and it should not be capped at 64 MB. New area sizes are set by JVM options: -Xmn<size> -Xmns<initialSize> -Xmnx<maxSize> A similar process can be used to set the size of HotSpot heaps. In addition to setting the minimum and maximum heap size, you should also increase the nursery size to between 1/4 and 1/2 of the heap size. Never increase the nursery to more than 1/2 the full heap. The nursery size is set using the MaxNewSize and NewSize parameters (that is, -XX:MaxNewSize=128m, -XX:NewSize=128m). After the heap sizes are set, use verbosegc traces to monitor GC activity. After analyzing the output, modify the heap settings accordingly. For example, if the percentage of time in GC is 78
IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch03.fm

high and the heap has grown to its maximum size, throughput might be improved by increasing the maximum heap size. As a rule of thumb, greater than 10% of the total time spent in GC is generally considered high. Increasing the maximum size of the Java heap might not always solve this type of problem as it is could be a memory over-usage problem. Conversely, if response times are too long due to GC pause times, decrease the heap size. If both problems are observed, an analysis of the application heap usage is required.

3.13.3 Setting the Heap Size when running multiple JVMs on one system
Each running Java program has a heap associated with it. Therefore, if you have a configuration where more than one Java program is running on a single physical system, setting the heap sizes appropriately is of particular importance. This is particularly true for 32 bit systems, where the total amount of addressable memory is limited. An example of one such configuration is when the IBM Integration Developer is on the same physical system as an IBM BPM Server utilizing a 32 bit JVM. Each of these is a separate Java program that has its own Java heap. If the sum of all of the virtual memory usage (including both Java Heaps as well as all other virtual memory allocations) exceeds the size of addressable physical memory, on the system, the Java heaps are subject to paging. This causes total system performance to degrade significantly. To minimize the possibility of this occurring, use the following guidelines: Collect a verbosegc trace for each running JVM. Based on the verbosegc trace output, set the initial heap size to a relatively low value. For example, assume that the verbosegc trace output shows that the heap size grows quickly to 256 MB, and then grows more slowly to 400 MB and stabilizes at that point. Based on this, set the initial heap size to 256 MB (-Xms256m). Based on the verbosegc trace output, set the maximum heap size appropriately. Care must be taken not to set this value too low, or Out Of Memory errors occur. The maximum heap size must be large enough to allow for peak throughput. Using the same example, a maximum heap size of 768 MB might be appropriate (-Xmx768m). This is to give the Java heap room to expand beyond its current size of 400 MB if required. The Java heap only grows if required (for example, if a period of peak activity drives a higher throughput rate), so setting the maximum heap size somewhat higher than current requirements is generally a good policy. Be careful not to set the heap sizes too low, or garbage collections will occur frequently, which might reduce throughput. Again, a verbosegc trace assists in determining this. A balance must be struck so that the heap sizes are large enough that garbage collections do not occur too often, while still ensuring that the heap sizes are not cumulatively so large as to cause the heap to page. This balancing act is configuration dependent.

3.13.4 Reduce or increase heap size if OutOfMemory errors occur

The java.lang.OutOfMemory exception is used by the JVM in a variety of circumstances, making it sometimes difficult to track down the source of the exception. There is no conclusive mechanism for telling the difference between these potential error sources, but a good start is to collect a trace using verbosegc. If the problem is a lack of memory in the heap, this is easily seen in this output. See 3.13.1, Monitoring garbage collection (GC) on page 77 for further information about verbosegc output. Many garbage collections that produce little free heap space generally occur preceding this exception. If this is the problem, increase the size of the heap. If there is enough free memory when the java.lang.OutofMemory exception is thrown, the next item to check is the finalizer count from the verbosegc (only the IBM JVM gives this
Chapter 3. Performance tuning and configuration

79

4784ch03.fm

Draft Document for Review January 27, 2012 7:40 am

information). If these appear high, a subtle effect might be occurring whereby resources outside the heap are held by objects within the heap and being cleaned by finalizers. Reducing the size of the heap can alleviate this situation, by increasing the frequency with which finalizers are run. In addition, examine your application to determine if the finalizers can be avoided or minimized. OutOfMemory errors can also occur for issues unrelated to JVM heap usage, such as running out of certain system resources. Examples of this include insufficient file handles or thread stack sizes that are too small. In some cases, you can tune the configuration to avoid running out of native heap. Try reducing the stack size for threads (the -Xss parameter). Deeply nested methods might force a thread stack overflow if there is insufficient stack size. For middleware products, if you are using an in-process version of the JDBC driver, it is usually possible to find an out-of-process driver that can have a significant effect on the native memory requirements. For example, you can use type 4 JDBC drivers (DB2's Net drivers, Oracle's Thin drivers), MQSeries can be switched from Bindings mode to Client mode, and so on. See the documentation for the products in question for more details.

3.14 Tuning for WebSphere InterChange Server migrated workloads

The tuning recommendations below are unique to workloads migrated using the WebSphere InterChange Server migration wizard in IBM Integration Developer. In addition to the recommendations specified, follow the other IBM BPM Server tuning recommendations detailed in this document: For JMS-based messaging used to communicate with legacy WBI adapters or custom adapters, make use of non-persistent queues when possible. For JMS-based messaging used to communicate with legacy WBI adapters or custom adapters, make use of WebSphere MQ-based queues if available. By default, the adapters use the MQ APIs to connect to the service integration bus destinations through MQ Link. MQ Link is a protocol translation layer that converts messages to and from MQ-based clients. By switching to WebSphere MQ-based queues, MQLink translation costs are eliminated and performance is improved. Turn off server logs for verbose workloads. Some workloads emit log entries for every transaction, causing constant disk writes reducing overall throughput. Turning off server logs may reduce the throughput degradation for such workloads.

80

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch04.fm

Chapter 4.

Initial configuration settings

In this chapter we recommend initial settings for several relevant parameters. These values are not optimal in all cases, but we have found that these values work well in our internal performance evaluations. They are, at a minimum, useful starting points for many proof-of-concepts and customer deployments. As discussed in 3.1, Performance tuning methodology on page 40, tuning is an iterative process. Follow that procedure and adjust these values as appropriate for your environment.

Copyright IBM Corp. 2012. All rights reserved.

81

4784ch04.fm

Draft Document for Review January 27, 2012 7:40 am

4.1 IBM BPM Server settings

This section provides settings based on IBMs internal performance evaluations of the IBM BPM 7.5.0 Server. These settings were derived using the tuning methodology and guidelines described in Chapter 3, Performance tuning and configuration on page 39. Consider these settings useful starting points for your use of this product. For settings that we do not list, use the default settings that are supplied by the product installer as a starting point, and follow the tuning methodology specified in 3.1, Performance tuning methodology on page 40. We discuss three settings in this section: A three-tier setup for BPEL business processes with the production database located on a separate server. A three-tier setup for BPMN business processes with the production database located on a separate server. A two-tier (client/server) setup for BPEL business processes with the production database co-located on the server.

4.1.1 Three tier configuration with Web services and a remote DB2 system (BPEL business processes)
A three-tier configuration was used in our internal performance work to evaluate the performance of a BPEL business process that models automobile insurance claims processing. This configuration is an example of many production environments, where DB2 is on a separate system than the IBM BPM Server. The Web services binding was used for communications. The business process has two modes of operation: A BPEL microflow (straight through process) that processed claims where no human intervention is required. A BPEL microflow plus macroflow (long running process) pattern, where the macroflow is invoked when a review or approval is required (for example, if the claim amount is above a certain limit). Three systems were used in this configuration: Request driver IBM BPM 7.5.0 server DB2 database server The IBM BPM server and the DB2 database server required extensive tuning to maximize throughput. Some tuning varied due to the operating system (such as AIX and Windows) and the number of processor cores. We present these variations in tabular format after the common tuning: The following settings and configuration options are recommended for all topologies in this section: Use the production template. Define the Common database as local DB2 type 4. Establish BPEL Business Process support with bpeconfig.jacl (This sets the Data sources BPEDataSourceDb2 WebSphere Application Server data source properties statement cache to 300). Disable PMI. Set HTTP maxPersistentRequests to -1.

82

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch04.fm

Set GC policy to Xgcpolicy:gencon (see Table 4-1 and Table on page 84 for nursery setting Xmn). Use remote DB2 databases (connection type 4) for BPE, SIB System, and SIB BPC databases. Table 4-1 shows IBM BPM 7.5.0 server-related settings to modify from their default value when the IBM BPM 7.5.0 server is deployed on AIX.
Table 4-1 Three tier application cluster settings for AIX Setting Java Heap Megabytes Java nursery Megabytes Xmn Default Thread Pool Max BPEDB Data source connection pool max BPEDB Data source WebSphere Application Server data source properties Statement cache size BPC messaging engine data source connection pool max SCA SYSTEM messaging engine data source connection pool max IBM BPM Common Data source connection pool max J2C activation specifications SOABenchBPELMod2_AS Custom properties maxConcurrency, maxBatchSize Resources Asynchronous Beans Work Managers BPENavigationWorkManager Work request queue size, max threads, growable Application Cluster Business Flow Manager Message pool size, Intertransaction cache size WebContainer Thread Pool Min,Max com.ibm.websphere.webservices.http.maxConnection Value 1536 768 100 300 300 50 50 500 50, 400, 50, no 5000, 400 100, 100 50

Table 4-2 shows IBM BPM 7.5.0 server-related settings to modify from their default value when IBM BPM 7.5.0 server is deployed on Windows and Linux on Intel systems.
Table 4-2 Three tier Web service and remote DB2 tuning variations Windows and Linux on Intel systems Tuning Variations Microflow: Number of Cores 1 Java Heap Megabytes Java nursery Megabytes -Xmn Web Container Thread Pool Max Default Thread Pool Max BPE database connection pool max 1280 640 100 100 150 2 1280 640 150 200 250 4 1280 640 150 200 250 Macroflow: Number of Cores 1 1280 768 100 100 150 4 1280 768 300 200 350

Chapter 4. Initial configuration settings

83

4784ch04.fm

Draft Document for Review January 27, 2012 7:40 am

Tuning Variations

Microflow: Number of Cores 1 2 30 40 80 40 4 30 40 80 40

Macroflow: Number of Cores 1 30 30 80 160 4 150 100 100 160

BPC messaging engine database connection pool max SYSTEM messaging engine database connection pool max Common database connection pool max J2C activation specifications SOABenchBPELMod2_AS Custom properties maxConcurrency BPEInternalActivationSpec batch size SOABenchBPELMod2_AS batch size Java custom property com.ibm.websphere.webservic es.http.maxConnection Application servers server1 Business Flow Manager allowPerformanceOptimizations Application servers server1 Business Flow Manager interTransactionCache.size Application servers server1 Business Flow Manager workManagerNavigation.messa gePoolSize Resources Asynchronous Beans Work Managers BPENavigationWorkManager min threads, max threads, request queue size

30 30 80 40

10 32 100 200 200 200

10 32 200

Yes

Yes

400

400

4000

4000

30, 30, 30

30, 30, 30

The DB2 database server has several databases defined for use by the IBM BPM 7.5.0 server. The database logs and tablespaces should be spread across a RAID array to distribute disk utilization. The SCA.SYSTEM.cellname.BUS database and the BPE database should be tuned as follows. The SCA.SYSTEM.cellname.BUS database should be tuned as follows: db2 update db cfg for sysdb using logbufsz 512 logfilsiz 8000 logprimary 20 logsecond 20 auto_runstats off db2 alter bufferpool ibmdefaultbp size 30000

84

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch04.fm

The BPE database should be created and tuned using the following DB2 commands and generated scripts:: db2 CREATE DATABASE bpedb ON /raid USING CODESET UTF-8 TERRITORY en-us

db2 update db cfg for bpedb using logbufsz 512 logfilsiz 10000 logprimary 20 logsecond 10 auto_runstats off Use the IBM BPM 7.5.0 server generated script: db2 -tf createTablespace.sql Use the IBM BPM 7.5.0 server generated script: db2 -tf createSchema.sql db2 alter bufferpool ibmdefaultbp size 132000 db2 alter bufferpool bpebp8k size 132000

4.1.2 Three tier configuration using Human Services with BPMN processes
A three-tier configuration was used in our internal performance work to evaluate the performance of a BPMN business process that models automobile insurance claims processing, using Human Services to process the claims with a Call Center scenario of Query Tasks, Claim Task, Complete Task, Commit Task. This configuration is an example of many production environments, where DB2 is on a separate system than the IBM BPM Server. The Web services Open SCA binding was used for communications. Three systems were used in this configuration: Request driver IBM BPM 7.5.0 server DB2 database server

BPM 7.5.0 Server

For a 64-bit JVM, alter Java memory management settings by adding the following Java command line parameters: -Xgencon, -Xms1800M, -Xmx1800M, -Xmn800M Increase size of the WebContainer ThreadPool to a minimum of 200, maximum of 400 Increase the maximum size of the Default Thread Pool to 40 Disable logging for selected BPM 7.5.0 components (e.g. Web Services)

DB2 Database Server

Separate log files and containers onto separate (RAID) disks Increase log file size for twproc database (to 16384 pages) Enable filesystem caching for twproc database: db2 alter tablespace userspace1 file system caching Exclude the table, SIBOWNER, from automatic runstats execution as described at: http://www-01.ibm.com/support/docview.wss?uid=swg21452323 Ensure that database statistics are up to date. If running at high throughput rates, consider disable some or all database auto maintenance tasks to avoid impacting peak throughput. However, if you disable these capabilities, be sure to perform runstats regularly to update database statistics Create new indexes as described in the technote at: http://www-01.ibm.com/support/docview.wss?uid=swg21474536

Chapter 4. Initial configuration settings

85

4784ch04.fm

Draft Document for Review January 27, 2012 7:40 am

Create the indexes using the following commands (if using DB2; other databases have similar commands): db2 "CREATE INDEX IDX_SOAB_BPD_INSTANCE ON LSW_BPD_INSTANCE (SNAPSHOT_ID ASC, BPD_INSTANCE_ID ASC) ALLOW REVERSE SCANS COLLECT SAMPLED DETAILED STATISTICS db2 "CREATE INDEX IDX_SOAB_BPD_INSTANCE_VAR ON LSW_BPD_INSTANCE_VARIABLES (BPD_INSTANCE_ID ASC, VARIABLE_TYPE ASC, ALIAS ASC) ALLOW REVERSE SCANS COLLECT SAMPLED DETAILED STATISTICS"

4.1.3 Two tier configuration using file store for JMS for BPEL business processes
This configuration was used in our internal performance work to evaluate the performance of a long-running business process that models a typical mortgage application process. A two tier configuration is used, with the IBM BPM 7.5.0 server and DB2 on the same physical system, as is common for proofs of concept when limited hardware is available. This configuration is not a representative production configuration, however. The JMS binding is used for communication. In this configuration, the BPE uses a DB2 database, and the messaging engines are configured to use file stores. To select the file store option, start the Profile Management Tool, select Advanced Profile Creation, and, on the Database Configuration window, select Use a file store for Messaging Engines (MEs). Tuning parameter settings for the BPE database were initially derived using the DB2 Configuration Advisor. A few key parameter settings are modified further: MAXAPPLS, which must be large enough to accommodate connections from all possible JDBC Connection Pool threads. The default buffer pool sizes (number of 4 KB pages in IBMDEFAULTBP) for each database are set so that each pool is 256 MB in size. Table 4-3 shows the parameter settings recommended for this configuration.
Table 4-3 Two-tier configuration using JMS file store parameter settings Parameter names APP_CTL_HEAP_SZ APPGROUP_MEM_SZ CATALOGCACHE_SZ CHNGPGS_THRESH DBHEAP LOCKLIST LOCKTIMEOUT LOGBUFSZ LOGFILSIZ LOGPRIMARY LOGSECOND MAXAPPLS BPEDB settings 144 13001 521 55 600 500 30 245 1024 11 10 90

86

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784ch04.fm

Parameter names MAXLOCKS MINCOMMIT NUM_IOCLEANERS NUM_IOSERVERS PCKCACHESZ SOFTMAX SORTHEAP STMTHEAP DFT_DEGREE DFT_PREFETCH_SZ UTIL_HEAP_SZ IMBDEFAULTBP

BPEDB settings 57 1 6 10 915 440 228 2048 1 32 11663 65536

In addition to these database-level parameter settings, several other parameters should also be modified using the administrative console. These settings primarily affect concurrency (thread settings): The amount of expected concurrency influences the size of the thread pool, because more in-flight transactions require more threads. For example, the size of the default thread pool might need to be increased beyond the default of 50 threads. The maximum concurrency is set to 50 threads for activation specifications. The database connection pool size for the BPEDB is increased to 60, and the statement cache size for the BPEDB is increased to 300. The Maximum Connections property for JMS connection pools is set to 40. Connectivity to the local database uses the DB2 JDBC Universal Driver Type 2 driver. Type 2 drivers produce better performance when the IBM BPM 7.5.0 server and DB2 are on the same physical system. IBM BPM 7.5.0 server JVM heap size is set to a fixed size of 1024 MB. In addition, the gencon garbage collection policy is used.

4.2 WebSphere ESB settings

This section discusses settings that are used for select internal performance evaluations of WebSphere ESB and were derived using the tuning methodology and guidelines described in Chapter 3, Performance tuning and configuration on page 39. Consider these settings useful starting points for your use of this product. For settings that are not listed, use the default settings supplied by the product installer as a starting point. See 3.1, Performance tuning methodology on page 40 for a discussion of tuning methodology.

4.2.1 WebSphere ESB common settings

These settings are good starting points, regardless of binding choices: Tracing is disabled.

Chapter 4. Initial configuration settings

87

4784ch04.fm

Draft Document for Review January 27, 2012 7:40 am

If security is required, use application security instead of Java2 security to reduce overhead. Java Heap size is fixed at 1280 MB for Windows and 1280 MB for AIX. Gencon garbage collection policy enabled (-Xgcpolicy:gencon), setting the nursery heap size to 1024 MB.

4.2.2 WebSphere ESB settings for Web services

The WebSphere ESB settings for Web services are as follows: PMI monitoring is disabled WebContainer Thread pool sizes set to max 50 and min 10 WebContainer Thread pool inactivity timeouts for thread pools set to 3500

4.2.3 WebSphere ESB settings for JMS

The WebSphere ESB settings for MQ and JMS are as follows: Activation specification Set maximum concurrent endpoints to 50 Queue Connection factory Set the maximum connection pool size to 51 DiscardableDataBufferSize Set to 10MB and CachedDataBufferSize set to 40MB

4.2.4 DB2 settings for JMS persistent

These settings are only relevant for JMS persistent configurations, as they make use of the database to persist messages: Place Database Tablespaces and Logs on a Fast Disk Subsystem Place Logs on Separate Device from Table Spaces Set Buffer Pool Size Correctly Set the Connection Min and Max to 30 Set the Statement cache size to 40 Raw partition for DB2 logs Otherwise, unless specifically noted in the workload description, the default settings as supplied by the product installer should be used.

88

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

4784bibl.fm

Related publications
The publications listed in this section are considered particularly suitable for a more detailed discussion of the topics covered in this paper.

IBM Redbooks
For information about ordering these publications, see How to get Redbooks on page 90. Note that some of the documents referenced here might be available in softcopy only. IBM Business Process Manager V7.5. Production Topologies, SG24-7976: http://www.redbooks.ibm.com/abstracts/sg247976.html

Online resources
These Web sites are also relevant as further information sources: WebSphere Application Server Performance URL http://www.ibm.com/software/webservers/appserv/was/performance.html WebSphere Application Server information center (including Tuning Guide) http://www-306.ibm.com/software/webservers/appserv/was/library/?S_CMP=rnav DB2 Version 9 best practices http://www.ibm.com/developerworks/data/bestpractices/?&S_TACT=105AGX11&S_CM DB2 Version 9.7 Info Center http://publib.boulder.ibm.com/infocenter/db2luw/v9r7/index.jsp IBM Java 6.0 Diagnostic Guide http://publib.boulder.ibm.com/infocenter/javasdk/v6r0/index.jsp The IBM Pattern Modeling and Analysis Tool (PMAT) for Java Garbage Collector http://www.alphaworks.ibm.com/tech/pmat Oracle 11g documentation library http://www.oracle.com/pls/db111/homepage IBM Integration Designer 7.5 Information Center http://publib.boulder.ibm.com/infocenter/esbsoa/wesbv7r5/topic/com.ibm.wbpm.wid .main.doc/welcome.html IBM Business Process Management 7.5 Information Center http://publib.boulder.ibm.com/infocenter/dmndhelp/v7r5mx/topic/com.ibm.wbpm.mai n.do.homepage-bpm.html Performance Tuning Resources for WPS and BPM solutions http://www.ibm.com/developerworks/websphere/library/techarticles/1111_herrmann/ 1111_herrmann.html

Copyright IBM Corp. 2012. All rights reserved.

89

4784bibl.fm

Draft Document for Review January 27, 2012 7:40 am

DB2 Best Practices for Linux, UNIX, and Windows http://www.ibm.com/developerworks/data/bestpractices/?&S_TACT=105AGX11&S_CMP=FP DB2 Version 9.5 Info Center http://publib.boulder.ibm.com/infocenter/db2luw/v9r5/index.jsp Using JCA Adapters with WebSphere Process Server and WebSphere ESB http://www-128.ibm.com/developerworks/library/ws-soa-j2caadapter/index.html?ca= drsWebSphere ESB Support http://www-306.ibm.com/software/integration/wsesb/support/ Oracle Architecture and Tuning on AIX v1.31: http://www-03.ibm.com/support/techdocs/atsmastr.nsf/WebIndex/WP100883 Oracle 10g Release 2 documentation (includes a Performance Tuning Guide) http://www.oracle.com/pls/db102/homepage

How to get Redbooks

You can search for, view, or download Redbooks, Redpapers, Technotes, draft publications and Additional materials, as well as order hardcopy Redbooks publications, at this Web site: ibm.com/redbooks

Help from IBM

IBM Support and downloads ibm.com/support IBM Global Services ibm.com/services

90

IBM Business Process Management (BPM) V7.5 Performance Tuning

Draft Document for Review January 27, 2012 7:40 am

Back cover

IBM Business Process Manager (BPM) 7.5 Performance Tuning and Best Practices

Redpaper
Learn valuable tips for tuning Get the latest best practices Try the example settings
This IBM Redpaper publication was produced by the IBM Business Process Manager (BPM) performance team. It provides performance tuning tips and best practices for IBM BPM 7.5 (all editions) and Business Monitor 7.5. These products represent an integrated development and runtime environment based on a key set of service-oriented architecture (SOA) and business process management (BPM) technologies: Service Component Architecture (SCA), Service Data Object (SDO), Business Process Execution Language for Web Services (BPEL), and Business Processing Modeling Notation (BPMN). This paper is aimed at a wide variety of groups, both within IBM (development, services, technical sales, and so forth) and customers. For those who are either considering or are in the early stages of implementing a solution incorporating these products, this document should prove a useful reference, both in terms of best practices during application development and deployment, and as a reference for setup, tuning, and configuration information. This paper provides a useful introduction to many of the issues influencing each product's performance, and can serve as a guide for making rational first choices in terms of configuration and performance settings. Similarly, those who have already implemented a solution using these products might use the information presented here to gain insight as to how their overall integrated solution performance might be improved. All of these products build on the core capabilities of the WebSphere Application Server infrastructure, so BPM solutions also benefit from tuning, configuration, and best practices information for WebSphere Application Server and the corresponding platform Java Virtual Machines (JVMs). Pointers to this information can be found in Related publications on page 89. The reader is encouraged to use this paper in conjunction with these references..

INTERNATIONAL TECHNICAL SUPPORT ORGANIZATION

BUILDING TECHNICAL INFORMATION BASED ON PRACTICAL EXPERIENCE

IBM Redbooks are developed by the IBM International Technical Support Organization. Experts from IBM, Customers and Partners from around the world create timely technical information based on realistic scenarios. Specific recommendations are provided to help you implement IT solutions more effectively in your environment.

For more information: ibm.com/redbooks

REDP-4784-00