>That's not what I meant. >Once the analysis, the understanding, and the gut >instinct have all been factored in, what numbers from IBM are meaningful to >use to specify to management which CPU will work and which one won't? >Given >that MIPS ratings are meaningless, how does one talk to management about how a >new CPU will handle the workload differently? >What's the basis for discussion?
I'm sorry for being unclear. As I said, I can't compress 30 years into a message, or two. The direct answer is it's not easy, and what I said is what you get. The evaluation, as I said in another post, is at best a guess. This has been an issue for over 30 years, and acknowleged for about 20. Even LSPR/zPCR has issues. There is no 'you can get there from here'. There is just experience (and sacrifices on the eve of the new moon). I'm not trying to duck, or make it sound like magic, but the bottom line is you do the best you can with the information you can gather. I've had many experiences, in 30 years, where we were wrong in estimating the capacity required. One, where we thought an R25 would be enought for a year, and we upgraded to an R35 within a week, and were at an R65 9 months later. Another, under COD, we did a 'temporary' upgrade and neve downgraded it. I could show you my scars (and my performance appraisals). When we agreed on the proposal, it was management and staff together, as a team. When it wasn't adequate, it was staff who had made the reco, and we were the ones who had fouled up. Mind you, fouled wasn't the word my VP used. - Ted MacNEIL [email protected] ---------------------------------------------------------------------- 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
