Selecting Job Steps to Monitor
Before you can begin to evaluate application performance, you must select which job steps that you want to monitor. You have three methods to make your selections:
- User-specified selection — You can designate the job steps that should be monitored by entering a monitoring request.
- System Management Facility (SMF) Candidate Utility PreLoader — You can run a utility that processes SMF data and places job steps showing high resource usage on a monitoring candidate list. From this list, you can select the steps that you want to monitor.
- Candidate AutoLoader — The AutoLoader will add job steps observed to exceed a set of thresholds to the monitoring candidate list. Then you can select them as additional job steps you want to monitor.
If you use the Candidate PreLoader or AutoLoader, you can define what job steps are eligible for the monitoring candidate list by specifying thresholds as criteria to make the list. Tailoring Generated Candidate Lists describes this function.
User-Specified Steps to Monitor
You can submit a monitoring request anytime you suspect a job step requires one. Basis data collection will occur until the job step has been monitored ten times. On subsequent executions, monitoring of the step continues to check for abnormal behavior. See Adding a Monitoring Request for more information about specifying AutoStrobe measurement requests on a step-by-step basis.
Generated Lists of Job Steps to Monitor
Two tools help you build a monitoring candidate list. You can first use the SMF Candidate PreLoader to scan SMF data to find monitoring candidates. Then to supplement what the PreLoader determined to be candidates, the AutoLoader will continue to survey job steps and add candidates to the list as they are found.
SMF Candidate PreLoader Utility Candidate Lists
This tool examines SMF information such as the frequency an application runs or number of EXCPS accumulated during program execution and builds a candidate list of job steps with resource usage that exceeds designated threshold values. See the section called SMF Candidate PreLoader Utility for more information.
AutoLoader Candidate Lists
Once you have created an initial monitoring candidate list by running the PreLoader utility, the AutoLoader will continue to track performance. If it observes high resource consumption by any job step, that step will be added to the monitoring candidate list. See Managing-AutoStrobe-Batch-Candidate-Processing-and-Monitoring for information about managing candidates after the AutoLoader has added them to the monitoring list.
Tailoring Generated Candidate Lists
Both the SMF Candidate PreLoader and the AutoLoader use threshold parameter values for resource usage to determine if a job step is eligible for the monitor candidate list. Existing default threshold values are set, but these can be overridden for both the PreLoader and the AutoLoader. You have the control of what job steps should be excluded from the candidate list by name, regardless of their performance.
Excluding Job Steps from the Candidate List
The exclusion list allows you to exclude applications that should never be a candidate for monitoring. System functions such as IEBGENER and other applications that are inappropriate can be placed on the exclusion list and will not be considered for the candidate list. See the section called Candidate Exclusion List for more information about building an exclusion list.
Thresholds for Candidate List Eligibility
The SMF Candidate PreLoader Utility adds data for a job step from the SMF data sets to the monitor-candidate lists for resource usage only if that step’s usage of one of the monitored resources met a “minimum threshold”. The PreLoader uses the values that are designated in the Strobe parameter library for these thresholds. The resources are:
- EXCPS
- TCB Time
- Elapsed Time
- Frequency of Run (only for the SMF PreLoader).
When Strobe is installed, parameters can be specified to change these values in the parameter library. If you want to change the value used by the PreLoader to ones that differ from the parameter library, specify new ones on the Strobe - AutoStrobe SMF Candidate Utility panel.
Similarly, Strobe has a default value for the number of times per week a job step must run to be eligible for the “frequency-of-run” candidate list. Strobe's default value for frequency is 30 executions per week. This parameter, AUTOCANFREQ, can be changed in the Strobe parameter library. However, this parameter value will always be overridden if you specify a new value on the Strobe - AutoStrobe SMF Candidate Utility panel.
The next section outlines several considerations for you to make about adding monitoring requests.