The behavior Ron describes below seems to fit what is happening. By the time I left work on Tuesday, we were nearing 20% full on the 4 local page datasets. As of this morning, the percent full is at 20%, so we seem to have reached a point where the page datasets stay at 20% full. The individual DB2 regions are holding pretty steady also.
This is certainly better than the 50 to 60% full the page datasets were before I redefined them to full pack 3390 Mod 3s. Eric On Wed, 12 Nov 2008 09:04:06 -0800, Ron Hawkins <[EMAIL PROTECTED]> wrote: >Eric, > >I don't think that this is unusual behavior. > >My SWAG on this is that the less busy pages in DB2, and other addresses, are being paged out during overnight batch, and then paged in again the next day when online becomes active. > >At this point you end up with is a copy of unchanged pages in storage and still out on AUX. until the pages in CS are changed, or pages on AUX overwritten you have two copies which allow page out/in to operate by a "bit-flick" rather than actual page movement. > >The jump to 20% is probably you working set. There will probably be a small jump each following day as different pages are stolen, but not the order of magnitude you see on the first day. > >Caveat Emptor - it's still a SWAG. > >Ron > Eric Bielefeld Systems Programmer Washington University St Louis, Missouri 314-935-3418 ---------------------------------------------------------------------- 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
