>And the MIPS tables have been pissing me off for some time. Modern workloads >are so complex these numbers really are pretty useless.
More than. I cannot tell everything I know, but: 1. They didn't test much of the big stuff. Rather, they produced a straight line from 17+ (z/990) and 21+ (z9) based on the 'MIPS', which is not linear. The degradation on everything greater than 2 CP's is linear. IE: if the third gives you 84% of the single engine, and the 4th gives you 81%, or whatever the figures are, the rest is a simple linear equation with a negative slope. Translating back to MIPS drops 3000 off the rated capacity of a 332. It's worse on the S54. I haven't gone public with the exact figures, but any body who can do the math, should take a close look at the LSPR figures. The actual point where they straightened the curve has changed a few times in the last couple of years. But, if you plot the (publicly) available figures, you'll see it. 2. They didn't test all the z/900 workloads on the z/990 (TSO), and they introduced new workloads (WebSphere) on the z/990 that they never ran on the z/900. How do you calibrate? And, PCR tells you what you can expect when you move WebSphere up (or even TSO)! How do they know? The algorithms may be valid; the input sure isn't! 3. With the introduction of Marketting MSUs, how are we going to be able to manage our bills easily? It was bad enough when we had to manage billing table for ISV's! Now, depending on the age of the box, we now have multiple tables for IBM. No wonder mainframe cost management is giving the plaform a bad name/perception! - -teD O-KAY! BLUE! JAYS! Let's PLAY! BALL! ---------------------------------------------------------------------- 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
