Note: Pages don't HAVE UICs anymore. They haven't for a long time. Just whether "referenced since last sweep through" or not.
Cheers, Martin Martin Packer, zChampion, Principal Systems Investigator, Worldwide Cloud & Systems Performance, IBM +44-7802-245-584 email: [email protected] Twitter / Facebook IDs: MartinPacker Blog: https://www.ibm.com/developerworks/mydeveloperworks/blogs/MartinPacker Podcast Series (With Marna Walle): https://developer.ibm.com/tv/mpt/ or https://itunes.apple.com/gb/podcast/mainframe-performance-topics/id1127943573?mt=2 From: "Joel C. Ewing" <[email protected]> To: [email protected] Date: 22/08/2016 14:58 Subject: Re: Low UIC count on z/OS 2.2 Sent by: IBM Mainframe Discussion List <[email protected]> On 08/22/2016 07:42 AM, Jaco Kruger wrote: > Good day, > > We are preparing to migrate from z/OS 1.13 to z/OS 2.2. The TST LPAR has 8GB Central Storage assigned and it resides on IBM z13. > > On z/OS 1.13, the UIC count is 65535 during and after the IPL. On z/OS 2.2, the UIC count is below 1000 during the IPL and it gradually increases. > > The Central Storage have been increased to 50GB on the TST LPAR, but the UIC count during the IPL stays below 1000, even when only basic components are started. There is no PAGING to Auxiliary storage. > > HVCOMMON, HVSHARE are DEFAULT values. LFAREA is set to 0M > > The WSET and Memory Object sizes (as obtained from RMF) have been compared between z/OS 1.13 and z/OS 2.2 and there are no candidates that justifies the low UIC count. > > Any idea on how to determine the reason for the low UIC count ? > > Regards > > Jaco Kruger > > It sounds like z/OS 1.13 was setting UIC of pages referenced during the IPL at unsigned-halfword high-values, which is really counter to the definition of UIC (seconds since page last referenced), and that z/OS 2.2 is more properly initializing UIC of IPL-related pages to zero, which would mean no page's UIC could possibly exceed 1000 until 1000 seconds (16.7 minutes) after the IPL. Perhaps initialization with high-values was an earlier convention to insure that physical pages which possibly had one-time use during the IPL process would always be the first to be reclaimed when production address spaces began to run and the available frame count became low, but I would suspect this could also force more LPA pages to be paged out that might subsequently have to be paged back in once production workloads were fully active. I would guess that with the much larger physical memory sizes in use today, any shortages of free pages occurs enough later after the IPL for normal page UIC value conventions to be sufficient to differentiate what pages are inactive without special logic to give unique treatment to pages touched during the IPL. That you are seeing a lack of any paging would seem to support this hypothesis, and that the low UIC seen with z/OS 2.2 is actually a fix, not a problem. Joel C. Ewing -- Joel C. Ewing, Bentonville, AR [email protected] ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN Unless stated otherwise above: IBM United Kingdom Limited - Registered in England and Wales with number 741598. Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
