> >>One of the most interesting points that he brought up was the > >>CPU 100 percent busy. > >>We all know that is not necessarily bad. I am not a perf cap > >>person but am reasonably comfortable with running my system(s) > >>at that . Now 20 year ago that was not the case but in all > >>reasonably current MVS systems that is a reasonable thing to do.
Not really. In order to get that famous 99.999% availability with a rock-solid SLA you need to run a data-sharing parallel sysplex -and- you need to leave some whitespace on each image so there is "room" to accommodate workload shifts for planned and unplanned system/application outages. If you don't do that and you lose an image (for whatever reason) you have a user perceived outage/loss of service and your 99.999 number goes poof or your SLA does. And if you're not in a parallel sysplex with data sharing you're never going to get close to those sort of numbers no matter what you do, even if you can make your SLA most of the time. > > Well, some are proud to be able to use all resources they paid > > for, others are proud to only use 60% of what they paid for and > > are happy they're wasting 40% ;-) I guess I'd be one of the "happy" ones wasting 40% Well, maybe not that much, but certainly a healthy chunk. A casual estimate for an n-way plex would be a bit less than 1/n spared per image, assuming you wanted the work of the failing/spared system to be picked up uniformly by the others. > And again - we hope, the Samsung mainframe was horribly oversized. Or > they will need much more HP machines. Or they will have strong > performance & security problems. > I simply don't believe that 7000MIPS datacenter grew up without real > need (workload). Did you read the Samsung guy's note? Samsung is a genuinely large business with real workloads. They made their decision to get off their mainframe for business reasons and it is a done deal. Their machines (plural) are sold, so it is safe to assume that whatever they're doing, they're really doing it on another platform (UNIX) just like they said they were. And apparently they're happier with the result. Their world did not come crashing down. We don't and can't know whether their mainframe systems were properly sized and effectively organized and run, but those questions are moot. That ought to be a cold shot of reality. Despite the mainframe's capabilities, there are significant downsides and not everyone is happy to accommodate those. As uncomfortable as it may be, the z ain't the king of the hill any more. CC ---------------------------------------------------------------------- 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
