They don't really want to know MIPS. They want to know money. That is, how much does it cost to run transaction x.
Perhaps expressing the CPU consumption as a percentage of the box's capacity would answer the question. Each engine has a fixed capacity of one CPU second per clock second, but the MIPS will vary depending on the speed of the engine (among other things). If, for example, you run a four engine, 100 MIP box, then one CPU second = 25% of the box or 25 MIPS. Another example is a four engine 1000 MIP box, then one CPU second is still 25% of the box or 250 MIPS. Another way to express the capacity of a box is the theoretical maximum number of CPU seconds you can use. Again assuming that you have a four engine box, then the maximum number you can use is 345,600. (24 hours x 60 min x 60 sec x 4). If this is a 100 MIP box, then each MIP is theoretically 34,560 CPU seconds. You should be able to see the problems that arise trying to express CPU seconds in MIPS. It get worse. We also have to consider the time value of that transaction. It may take more than one MIP to run 34,500 one second transactions with an acceptable response time. You should be able to see the problems that arise trying to express CPU seconds in MIPS. Remember what we are trying to do and who is asking. I'd offer the second calculation and see what they say. Remember, no jargon and KISS when framing your reply. HTH and good luck. -----Original Message----- From: IBM Mainframe Discussion List [mailto:[email protected]] On Behalf Of Ramiro Camposagrado Sent: Thursday, July 02, 2009 10:26 AM To: [email protected] Subject: Converting CPU Time to MIPS Management is trying to cut and save CPU cost. So they came up with this "Cost Avoidance Committee". (Primarily consisting of uper management). Now they want us to somehow equate the CPU Time consumed (primarily by a CICS transaction) and equate that into MIPS. I know that folks have come up with formulas on how to do this. We use BEST/1 (Visualizer) for our performance reporting and it works off 15 minute intervals. Within this 15 minute interval, we can run hundreds of CICS transactions. If we take one CICS transaction and convert the CPU time used into MIPS, it is much higher than the 15 minute interval that is being reported in BEST/1. So there is really no way for us to correlate this number with our BEST/1 data. The performance data comes from MXG and it gets fed into BEST/1. As far as I know, MXG does not report on MIP utilization per CICS transaction. It does report on the elapsed time and CPU time for individual CPU transactions. And it would be just to much of an effort for us to report on "MIPS used" for hundreds of CICS transactions. Any suggestions or comments ?? Regards, Ramiro ---------------------------------------------------------------------- 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 NOTICE: This electronic mail message and any files transmitted with it are intended exclusively for the individual or entity to which it is addressed. The message, together with any attachment, may contain confidential and/or privileged information. Any unauthorized review, use, printing, saving, copying, disclosure or distribution is strictly prohibited. If you have received this message in error, please immediately advise the sender by reply email and delete all copies. ---------------------------------------------------------------------- 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
