On Mon, 25 Sep 2006 15:14:02 +0000, Ted MacNEIL <[EMAIL PROTECTED]> wrote:
>>>Perhaps it's the design of WLM that doesn't account >>>for profoundly diverse behaviors within a single LPAR. > >>Ahh, but it does. > >THAT I disagree with! > >Too many times have I found that mixing I/O bound and CPU bound jobs >on the same system kills the performance of the I/O bound jobs. Are you seriously suggesting that your workloads will run better if you put the I/O bound work in a different LPAR than the CPU bound work? > >The only goal that has MTTW (wherein this can be detected and >compensated for) is discretionary. Yes, and IMHO, WLM works best if there are discretionary workloads > >I have never had the luxury of having a machine idle enough to allow >that workload to even run. Does all your work get done? If so, putting some in discretionary will not reduce the total amount of work that gets done. > >We are a world-wide country with workloads peaking all day long. >In that environment, we can hardly get even the 4's & 5's to run. Importance 4 & 5? You can't get your most important work to run? Why not? > >The answer is an upgrade; the cost is an issue. Of course. Including the cost of work that doesn't finish on time. >(We are out-sourced, with all that entails) Including service level agreements? Tom Marchant ---------------------------------------------------------------------- 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
