>IIRC there is a statement in the SMF manual about CPU time not being >repeatable.
This has been the case since pre-MVS/XA. IBM documented this in many places over the last 25+ years, and gave explanations at seminars such as CMG. The simple explanation is, during one of the MVS/SP1.x releases, some things that were done in disabled mode, and under SRB reported CPU, were done in disabled mode and TCB which was allocated to the last active task. This was, in a way, to reduce overhead, and make the system more responsive. But, since some things were 'sensitive', whatever system related function was being done had to start over again when it was resumed. We discovered this, at an insurance company I was working at, when we were bench marking an upgrade/replacement of a 470 with a 3081. When we pointed out the discrepency to IBM, they tried to tell us our methodology was flawed, and our tools weren't up to the level of MVS we were running. We pointed out we were using SMF & RMF only, and showed that we were current on PTF's. When we threatened to go for a 5840, they opened up and explained the issue. The whole event soured me on bench marking. The overall day to day work of a busy corporate environment can be consistent. But, individual jobs are not. My understanding of VM, is that it introduces another level of variability, but I have never measured individual jobs under a guest operating system. I've just measured the entire guest, or collection of guests. - I'm a SuperHero with neither powers, nor motivation! Kimota! ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
