The amount of time required for the programs to process SMF data depends on such environmental considerations as the device type on which the SMF data resides, the processor speed, the operating system being used, and the nature of other work being run at the same time. The best estimate for your environment can probably be obtained by looking at run times for existing jobs that process SMF data in similar quantities (such as accounting routines).

We anticipate that CRURRSV will have considerably more data to process than CRURRAP and have therefore provided the following observations for optional use as a general guideline. These observations also show the benefit to be obtained from specifying additional buffers.

The main element in the run time is I/O activity. Therefore, if you are in a recovery situation and your objective is to minimize execution time with little consideration for virtual storage usage, you should specify additional buffers on the BUFNO DCB subparameter in the JCL for SMFIN and SMFOUT (CRURRSV), and for SMFIN, EXPIN, and EXPOUT (CRURRAP). Assuming cylinder allocation for DASD data sets, and availability of sufficient real storage, we suggest buffer specifications which will give around 60 buffers or 500K of buffers, whichever uses less virtual storage, for each of the DDnames above. This should minimize the run time. However, as you can see from the two previous examples, the run time is likely to be fairly short in any case.

The region size information in Specified Operating Environment assumes default buffering (5 buffers for each of the DDnames above). Increasing the number of buffers may require an increase in the region size.