Defining service classes and performance goals
A service class is a group of work with the same performance goals, resource requirements, or business importance. When you define a service class to WLM, you also assign performance goals to the service class.
Performance goals can be one of the following:
- Discretionary
- Work is run when the system resources are available.
- Response Time Goal
- A response time goal can be either an average response time or
a response time and a percentile. Response time goals are measured
from the time the job completes C/I processing (main service arrival
time) until it completes execution.
Conversion delay is still tracked by JES3, but it is not included as part of the goal. TYPRUN=HOLD time is not included in either the conversion delay or in the response time calculation.
- Velocity Goal
- Velocity is a measure of how fast work should run when ready, without being delayed for WLM-managed resources. For jobs in JES-managed job class groups, the definition of velocity is job initiation through job termination. For jobs in WLM-managed job class groups, velocity includes queue delay (the amount of time waiting for an initiator). By including queue delay in the velocity calculation, WLM can determine whether adding another initiator will help a service class meet its performance goals.
Figure 1 illustrates response time/velocity
goals.
Figure 1. Response
Time/Velocity Goals
* Input Service (Reader) <***********
- User Delay *
*
*
* Converter/Interpreter *
- Conversion delay *
*
*
* Main Service * <*********
- Operational delay (job hold, class disabled) * *
- JES scheduling delay (class limits, duplicate * *
job name) * Response *
- Resource delay (resource affinity, JES3 setup) * Time *
- Queue delay (waiting for an initiator) * <*********** *
* *
* Velocity WLM- *
* Initiation <************* * managed Job *
Velocity JES- * Class Groups *
managed Job * *
Class Groups * *
* Execution <************* <*********** <*********** <*********
* Output Service
* Purge