JMS@ARIZMIS.BITNET (Joel 'Bud' Headroom) (09/24/86)
>From: EDU%"A105%UWOCC1.BITNET@WISCVM.WISC.EDU" 23-SEP-1986 21:54 >Subj: VAX CPU usage not reproducible > Help again. It has been brought to my attention here that a user >running a fixed job can be charged varying amounts. Our charging algorithm >is simply rate/hour times CPU used. This means that a user performing the >same task repeatedly with identical data is using differing amounts of >CPU resource. For our site, this is not good. > My impression is that some (perhaps all?) system overhead is being >tallied against user jobs. Is this so? Is there any way to stop this >happening? For our site, reproducibility (ie fairness) is important. > One of the factors which specifically seems to be at fault is paging. >As available memory saturates, and paging increases, the individual job >costs rise. There may be other components affecting us as well. > If there is no way to remove system overhead from our cost calculations, >has anyone investigated ways of estimating it (based on paging for the job >and/or other factors)? If so, the estimated overhead could be applied to >the cost calculations to reduce the impact of the overhead. All I can say is, "gee, you don't like that behavior either?" I spent a whole month back in 1979 at VMS V1 trying to answer the questions above, and the answer then, and still is, and will have to be for future VMS versions, "No, there ain't no way to be fair about it!" This is one of my favorite 'bugs' in VMS; I hate to see someone bit by it like the University is, but there is no way to deal with it. The entire system philosophy would have to be changed to fix it, and there doesn't seem to be enough pressure on Digital to do that. Some hints, though : For a given working set, and assuming that your machine is not limited on memory, the page fault total is the most reproducible statistic you can get. We suggest combining connect time, CPU time, and page faults to get a reliable metric, but that assumes that all users have the same working set quotas AND that your system is not memory constrained (although this screws it up less than you might think). VMS charges you for what it does, and what it does takes a variable amount of time, and there's no way to get around it. The system simply cannot, and will not, ever be fair and reproducible. Even on a no-load system, the time is not really reproducible, since things like placement of files on disk can produce 5% changes or more. jms +-------------------------------+ | Joel M Snyder | BITNET: jms@arizmis.BITNET | Univ of Arizona Dep't of MIS | ArizoNET: MRSVAX::JMS | Tucson, Arizona 85721 | Pseudo-PhoneNET: (602) 621-2748 +-------------------------------+ (std. disclaimer in re: nobody taking anything I say seriously) --*> "Wherever you go ... there you are." -- Buckaroo Bonzai <*--