Observing bottlenecks and exceptions

There are two approaches to monitoring performance:

• You can check the performance of your system by observing indicators like performance goals, performance index, workflow, or response times, most of which have already been discussed.
• Or you can use a variety of Monitor III reports to look for exceptions or delays that might be the source of a performance problem.

To define your performance exceptions directly, use the

• Monitor III - Workflow/Exceptions Report: There are many types of exceptions you can specify: CPU utilization, response times, number of users, storage activities and many others.

  You define thresholds and corresponding colors, and Monitor III indicates when a threshold has been reached.

Now, you can start investigating the reason, which hopefully will enable you to solve the problem either immediately, or with the next IPL, or with planning for a more powerful processor in the near future.

If you would prefer information about exceptions for a longer time range, you will call the

• Postprocessor - Exception and Overview Reports: For these single-system reports, you can define many types of exceptions or thresholds, on the basis of CPU, I/O, workload or paging data gathered by Monitor I. The reports list all relevant data and allow you to create the detailed interval reports.

The other method is to look directly for delays — situations in which jobs are waiting for resources (for example, processor, devices, storage).

• Monitor III - System Information Report: This report gives you an overview of all applications in your system at different levels (system, TSO, batch, and so on) or grouped by workloads or service classes. The information includes speed and delay indicators.

If you like to create your own performance reports that should contain the data you are interested in, you can do this with RMF PM.

• RMF PM - PerfDesks: RMF PM takes its input data from Monitor III. The data is suitable for monitoring and analyzing performance in real time and in the recent past. It provides a selected subset of the information provided by the Monitor III gatherer: general performance data, performance data for jobs, and workload-related performance data.

  You can collect real-time data, combine data from different collection types, or even from different applications, and group resources together. Once you have created these scenarios, you can save them in your own panels, called PerfDesks.