Nguyen Duc wrote: >...doing some measurements i just noticed that there is a noticeable >difference between DB2 CPU times i got from SMF30, and the times i got from >the DB2 statistics records (SMF100) about the CPU consumption of DB2 address >spaces.
>With excel , the 2 graphs are parallel , so i guess that there is something >that is not accounted with the SMF100 figures. My MVS colleague just noticed >that there is only 2% difference in the total consumption of the day, and >think that it may come from the way values are consolidated (SMF interval vs. >DB2 stats interval) Hmmm, I would say 'overhead', but I will rather STFU! :-D (because I'm not a DB2 expert/specialist/guru! I do however some measurements for performance purposes.) Questions so other experts may help you: What versions of DB2 and z/OS do you have? Do your DB2 work with CICS for example? What subtypes of SMF 30 records are you using? What are your intervals from SMF, RMF and DB2? Can you drill down to see if you can see any patterns of transactions and type of transactions or who is issuing those transactions? Just curious about your transaction mix... Are those CPU values from batch jobs or interactive? Could you see any patterns from RMF as additional help? Perhaps you can post your questions on DB2-L. Groete / Greetings Elardus Engelbrecht ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
