One simple way to answer the original question, the CPU variability for a single program, under different load conditions, is to measure the actual variability.
Get your test system for an hour. Set the RMF and SMF intervals synchronized at 1 MINUTE. Read several hundred to several thousand copies of the job into the internal reader, all same class, with no inits started. Now, start 10 more inits every five minutes, (10% of the number of inits at 100% capacity) and you'll have driven the system load from nothing to full load. Now, plot the seven individual CPU time buckets and the total CPUTM for each step versus time, and you will see the magnitude of the variations, versus system load. You can also examine the minute by minute RMF data to see when this load saturated which resources. Barry Merrill ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
