Hi all, Yes, 30 minutes was my recommendation for safety. But if it were my shop, I would use 15 minutes, but not 5. Here’s my reasoning. My recommendations go out to people who are experienced and those who are newbies. When SMF writes its records, it’s one of the highest priority tasks. If there are too many records to write, it could tie up other tasks needing to free locks, etc. So using 30 minutes reduces those exposures.
But as several people mentioned, most of the peaks are hidden during 30-minute periods. So I would use 15 and make sure that nothing gets locked out during syncing. The reason I wouldn’t use 5 minutes is the same reason. I’ve seen too many times with a sync of 5 minutes where there is work that is blocked. My technique is to use an external monitor or RMF Monitor III at 1 minute intervals. The overhead is really minimal. We have been busy the past few years on CMP, pricing, and new SW/HW, so the latest info I have on SMF parameters was in 2015 in a SHARE presentation. You can see that at https://watsonwalker.s3.us-west-1.amazonaws.com/ww/wp-content/uploads/2016/01/28151211/PR150812a.pdf. Alternatively, you can go to https://watsonwalker.com/publications/presentations/ and scan down to 2015. Cheers, Cheryl On Dec 29, 2023, at 6:20 PM, Mark Zelden <m...@mzelden.com> wrote: On Fri, 29 Dec 2023 13:57:35 -0600, Mark Zelden <m...@mzelden.com <mailto:m...@mzelden.com>> wrote: > On Thu, 28 Dec 2023 21:23:18 -0800, Ed Jaffe <edja...@phoenixsoftware.com> > wrote: > >> What SMF interval do most folks use? >> > > In my experience (from many shops / clients over the years), it matches the > RMF interval > and the most common if 15 minutes. Second most common is probably 30 (along > with > RMF) but I think most shops moved away from that to go to at least 15 years > ago. I have > seen some use 5 minutes and sometimes IBM will request that for a period of > time - perhaps > for a week to get a more accurate picture for a CP3000 study. > > This is typically what I use in SMFPRMxx: > > INTVAL(15) > SYNCVAL(59) > You didn't way why you wanted to know. But thinking about this more... I though i remembered Cheryl Watson doing a poll on this once. I searched her website and saw in 2008 there was a 3 part series on SMF / parms and she asked people to send in their parms, but I didn't see a follow up on the results. She did recommend setting INTVAL(30) and said using SYNCVAL(59) was no longer required and to use SYNCVAL(0). I won't go into the history for why people started coded SYNVCAL(59) to begin with (she does). Maybe someone on team Cheryl does have poll results from back then or more recently. However she also recommended changing RMF invterval from 15 to 30 to match SMF INTVAL (she previously suggested using RMF interval of 15, SMF INTVAL(30). Partially due to the number of SMF/RMF Type 74 records from DASD activity from the size of systems at the time. That to me makes no sense because even though there is more RMF data to store and process, the CPUs are much faster, the disk & I/O is much faster and storage is "cheaper", so it's all relative. I know I'm talking RMF interval now as opposed to your question on SMF INTVAL, but 30 minutes is just not granular enough in the large installations I have worked in. Be it for typical performance report & capacity planning or looking at WLM reports (although I use RMF III or Mainview CMF more for WLM tuning that post processing). Even in small environments I have always used 15 for both SMF and RMF/CMF. Back to your question: While I have mostly seen 15 minutes to match RMF / CMF 15 minutes, in my personal experiences, 30 minutes is the default and lot of people listen to Cheryl's advise (because it is good), so without any scientific polling, I'm sure that it is still very common to see INTVAL(30). I just don't agree and have never used anything higher than 15. This paper from Scott Chapman of EPS talks about the subject and he agrees with me that it should be no longer than 15 minutes and that RMF/SMF should be synced. https://www.pivotor.com/library/content/Chapman_SMFRecommendations_2022.pdf Best Regards, Mark -- Mark Zelden - Zelden Consulting Services - z/OS, OS/390 and MVS ITIL v3 Foundation Certified mailto:m...@mzelden.com Mark's MVS Utilities: http://www.mzelden.com/mvsutil.html ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN Cheryl Watson ====================== Cheryl Watson Walker, CEO Watson & Walker, Inc. Sarasota, FL USA www.watsonwalker.com Cell/Text: 941-266-6609 ====================== ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
