Previous Topic: Functional OverviewNext Topic: Batch Tasks Replaced by Using the Server


Getting Information

In general, MVS provides information about the performance of batch jobs by using the following components:

RMF provides information related to the physical CPU environment, for example, processor utilization of address spaces, I/O wait times, and so on. Such information is presented at the address space level. Because a batch job can have more than one address space, there is usually no direct correlation between a specific batch job of interest and the information provided.

On the other hand, SMF provides information about the system activity, for example, started jobs, I/O requests issued from a job, and so on. SMF writes this information to SMF data sets, which can be later inspected offline with SAS or other user-written analysis procedures. SMF also writes the data collected by RMF to its own SMF data sets.

Analyzing SMF data offline is not desirable for controlling the performance of an MVS batch environment because the length of time between the analysis and the point of time where an action must be initiated is too long. The switching of SMF data sets is installation and activity dependent.

However, SMF provides a number of user-exits where performance control can hook into and analyze the SMF data at the point when they are created and before the SMF data is written out to the SMF data set. The user-exit candidates for performance control are as follows: