Workflow and delay monitoring
Workload workflow
Workload workflow measures how efficiently system resources are serving the jobs in a workload (batch, Started Tasks, or TSO sessions):
- A high workflow value indicates that the workload is accessing resources on request with little impact on performance because of resource contention.
- A low workflow value typically indicates that the workload is accessing few of the system resources it needs to execute.
Workload Workflow Calculation:
where:
- # samples using system resources is the number of samples collected that show jobs in the workload that are using processor and device resources.
- # samples delayed for system resources is the number of samples collected that show jobs in the workload that are delayed for resources.
Workload delay
Workload delay measures the impact on a workload’s performance because of contention for physical and logical resources.
Delays typically occur because resources are used sequentially; if a resource is currently servicing a job, other jobs that are requesting the same resources are delayed until the current request ends.
- A high delay value indicates that jobs in the workload cannot gain access to the resources they need.
- A low delay value indicates that system resources are serving the jobs in a workload efficiently, which indicates little contention for resources from other jobs.
In most cases, the sum of the values for workflow and delay is not 100%, because of voluntary wait or idle time (for example, TSO user address space think time, when the job is neither using resources nor delayed because of contention for a resource).
Workload Workflow Calculation:
where:
- # samples delayed for system resources is the number of samples collected that show jobs in the workload delayed for resources such as processor, DASD devices, storage, enqueue, SRM, or HSM.
- total # of samples is the total number of samples collected.
Interpreting delay in terms of response time
You can also statistically interpret workload delay in terms of time, which lets you evaluate the impact of delays.
For example, assume that average TSO response time is two seconds and that the delays observed for the TSO workload are in the percent column. You can quickly identify areas for improvement in terms of time, as shown in the following example:
Delay due to | Percent | Seconds delay |
|---|---|---|
processor | 45% | .9 |
devices | 30% | .6 |
storage | 10% | .2 |
SRM | 5% | .1 |
Job workflow
Job workflow indicates how effectively system resources serve a job:
- A high job workflow value implies that a job is accessing the requested resources, and other jobs have little impact on it.
- A job with a low workflow value has few of the resources it needs to execute.
Job workflow is calculated in a similar manner to workload workflow.
Job delay
Job delay is a measure of resource contention on a job’s performance:
- A high delay value suggests that the job cannot access the resources it needs because of contention from other jobs.
- A low delay value indicates that the job is experiencing little contention for resources from other jobs.