IBM BPM, V8.0.1, All platforms > Tuning

Performance tuning methodology

Use a system-wide approach when tuning performance of a Business Process Management environment. To provide acceptable performance, you should understand some key aspects of tuning.

Tuning encompasses every element of the deployment topology:

• Physical hardware topology choices
• Operating system parameters tuning
• IBM BPM, WebSphere Enterprise Service Bus, WebSphere Application Server, and DB2 for z/OS tuning

The methodology for tuning is iterative:

1. Select a set of reasonable initial parameter settings.

2. Run the system.
3. Monitor the system to obtain metrics that indicate whether performance is being limited.

4. Use monitoring data to guide further tuning changes.
5. Repeat until you are satisfied with the performance.

The following sections provide more details on some of the methodology steps.

Select a set of reasonable initial parameter settings:

Use the tuning checklist for a systematic way to set parameters.

Monitor the system to obtain metrics that indicate whether performance is being limited:

Monitor the system to determine system health and determine the need for further tuning. The following resources must be monitored:

• Each physical system in the topology, including front-end and back-end servers, such as web servers, and database servers

• Processor usage, memory usage, disk usage, and network usage using relevant tools for your platform
• Each JVM process started on a physical system, for example, IBM BPM controller, servant, and adjunct
• Verbosegc statistics

• For each physical machine in the topology (including front-end and back-end servers, such as web servers, and database servers), monitor CPU usage, memory usage, disk usage, and network usage using relevant z/OS tools, such as SMF and RMF™.
• Each JVM process started on a physical system, for example, the IBM BPM controller, servant, and adjunct:

  • Use tools, such as RMF Workload Activity reports to get CPU and memory usage per process.
  • Collect verbosegc statistics.

• For each IBM BPM server, use Tivoli Performance Viewer (TPV) or WebSphere Application Server Performance Monitoring Infrastructure (PMI) to monitor the following resources:

  • For each data source, the data connection pool usage

  • For each JCA resource connector, the connection and session pool statistics

  • For each thread, monitor usage

    This command is helpful for understanding thread usage on WebSphere Application Server for z/OS (for example, D OMVS,PID= xxxxxxxx)

Use monitoring data to guide further tuning changes:

Look at the collected monitoring data, detect performance bottlenecks, and do further tuning. This phase of tuning is driven by the monitoring data collected in the previous phase.

The following list contains examples of performance bottlenecks:

• Excessive use of physical resources, such as processor, disk, and memory. To resolve usage issues, add more physical resources, or rebalance the load more evenly across the available resources.
• Excessive use of virtual resources, such as heap memory, connection pools, and thread pools. To resolve usage issues, use tuning parameters to remove the bottlenecks.
• Inefficient DB2 use. Use DB2 for z/OS performance monitoring tools to monitor your database. For a list of tools available for monitoring DB2 for z/OS, go to the DB2 Version 9.1 for z/OS Information Center and search for Using tools to monitor performance.

Tuning