DB2 10.5 for Linux, UNIX, and Windows

Monitoring and intervention

The third domain of workload management is monitoring, which must be performed on an ongoing basis.

The primary purpose of monitoring is to validate the health and efficiency of your system and the individual workloads running on it. Using table functions, you can access real-time operational data such as a list of running workload occurrences and the activities running in a service class or average response times. Using event monitors you can capture detailed activity information and aggregate activity statistics for historical analysis.

Looking at aggregate information should usually be the first step when you build a monitoring strategy. Aggregates give a good picture of overall data server activity and are also cheaper because you do not have to collect information on every activity in which you might be interested. You can collect more detailed information as you understand the scope of your monitoring needs.

Typical monitoring tasks you can perform are:
  • Analyzing the workload on your system to help design your initial DB2® workload management configuration.
  • Tracking and investigating the behavior of your system by obtaining types of operational information that permit you to:
    • Analyze system performance degradation
    • Diagnose activities that are taking too long to complete
    • Investigate agent contention
    • Isolate poorly performing queries
    Information is available for activities, service classes, workloads, work classes, threshold queues, and threshold violations.
  • Exercising control over the execution environment by canceling queued activities that you expect will cause problems or cancel running activities that you have diagnosed as negatively impacting the system.