Allan, Paging rates depend on the interval they are measured over. A page rate of one or two per second measured in a 15 minute RMF interval may be due to a page rate of 30-60 pages/sec for half a minute within that RMF interval.
Back before the days of Expanded Storage I established the total page movement rate from One minute Monitor II intervals. It's surprising just how much larger the peaks are when you drop down to this small an interval. Ron > -----Original Message----- > From: IBM Mainframe Discussion List [mailto:ibm-m...@bama.ua.edu] On Behalf Of > Staller, Allan > Sent: Thursday, January 21, 2010 5:27 AM > To: IBM-MAIN@bama.ua.edu > Subject: Re: [IBM-MAIN] PAGE datasets -- few large or more small > > 1) 100% agreed > 2) 100% agreed > 3) 90% agreed. I agree with statement paging bandwidth... I disagree > with the second statement more smaller is always better... The > bandwidth must be adjusted to suit the paging rate. If the total page > rate is 1 or 2 per sec, one local could handle the load easily. If it is > 100's per sec, a more robust paging subsystem would be needed. > Converting 9 locals to 3 may or may not hurt! (it depends). > 4) If I had an unlimited budget.... > > <snip> > - Individual local page datasets should not exceed 30% > (or > 25% if you choose) utilization because the contigious slot allocation > algorithm becomes inefficient/unsuccessful for that dataset. The > utilization% at a total paging subsystem level is irrelevent with > respect to > the algorithms. > > - All local page datasets should be the same size (or > close > to it in slots) and device type because of ASMs round robin allocation > algorithm. If one page dataset has 12000 slots and another has 24000 > the > utilization% on the second will be about 1/2 of the first. > > - Paging bandwidth is as important as paging space. > Converting 9 page datasets to 3 page datasets of triple size will > probably > impact performance. More smaller page datasets is always better over > fewer > large page datasets. > > - A volume should only have a single local page dataset > on > it, and no other datasets. Use of WLM managed PAVs by ASM can be used > to > eliminate this restriction. Depending on the actual storage subsystem > backing a page dataset it may or may not be reasonable to have multiple > systems using a volume with a page dataset on it. > </snip> > > ---------------------------------------------------------------------- > For IBM-MAIN subscribe / signoff / archive access instructions, > send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO > Search the archives at http://bama.ua.edu/archives/ibm-main.html ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
