From: CRDGW2::CRDGW2::MRGATE::"SMTP::CRVAX.SRI.COM::RELAY-INFO-VAX" 2-NOV-1990 03:07:03.71 To: MRGATE::"ARISIA::EVERHART" CC: Subj: Re: MPW parameters suggestion. Received: by crdgw1.ge.com (5.57/GE 1.76) id AA15113; Fri, 2 Nov 90 02:48:39 EST Received: From UCBVAX.BERKELEY.EDU by CRVAX.SRI.COM with TCP; Thu, 1 NOV 90 19:47:55 PDT Received: by ucbvax.Berkeley.EDU (5.63/1.42) id AA01535; Thu, 1 Nov 90 19:36:43 -0800 Received: from USENET by ucbvax.Berkeley.EDU with netnews for info-vax@kl.sri.com (info-vax@kl.sri.com) (contact usenet@ucbvax.Berkeley.EDU if you have questions) Date: 1 Nov 90 13:53:29 GMT From: crash!orbit!incstar!lhotka@nosc.mil (Glamdring) Organization: INCSTAR Corp, Stillwater MN Subject: Re: MPW parameters suggestion. Message-Id: <932@incstar.uucp> References: <0FC980B7DF7F407825@iscsvax.uni.edu> Sender: info-vax-request@kl.sri.com To: info-vax@kl.sri.com In article <0FC980B7DF7F407825@iscsvax.uni.edu>, CONNELLY@ISCSVAX.UNI.EDU (Ken Connelly - VAX System Manager) writes: > In message: <11383E66FAFF003CC9@ewuvms.bitnet>, Dennis Helsing > says: >> MPW_HILIMIT = 15% of the total physical memory size. >> MPW_WAITLIMIT = MPW_HILIMIT >> MPW_LOLIMIT = half of MPW_HILIMIT >> MPW_THRESH = 16,384 >> >> Our 64mb is divied up as follows... >> MPW_HILIMIT = 19,660 >> MPW_WAITLIMIT = 19,660 >> MPW_LOLIMIT = 9,830 >> MPW_THRESH = 16,384 >> > What is the general consensus of the group as to these "rules of thumb"? > We also have a 6410, with 128MB RAM. VPA has been telling me fairly > [VPA results deleted...] We have an 8350 with 54 megs (soon to be a 6410 with 128 megs), and we use pretty much those parameters with a couple exceptions. I too can not take credit for the following information, as I got it via DECUS seminars as well... (Credit goes to Technology Solutions Incorporated, (c) 1989) MPW_HILIMIT should be 15% of physical memory OR 15000 whichever is LOWER. The reasoning behind this is to prevent the modified page list from growing disproportionatly large. MPW_THRESH = 65535 This is set high to prevent the modified list from being flushed if at all possible. The 16384 value is good for VMS 4.x, but should be increased for 5.x. MPW_WAITLIMIT = 65535 This is set high to prevent processes from ever begin held up by modified list writing. The 16384 value is good for VMS 4.x, but should be increased for 5.x. For VMS 5.x the following also apply: MPW_LOWAITLIMIT = MPW_HILIMIT - (2 * MPW_WRTCLUSTER) MPW_IOLIMIT = at least 8. Can be larger if you have fast disks/more page files. Other critical SYSGEN parameters to watch are PAGEDYN, NPAGEDYN and the SRP*, LRP* and IRP* parameters. Any of these which are too small will cause serious problems as well. During peak usage periods do a SHOW MEMORY and examine these parameters: Nonpaged Dynamic Memory (NPAGEDYN) should be at least 100,000 Paged Dynamic Memory (PAGEDYN) should be at least 200,000 SRP List should have 100 free IRP List should have 50 free LRP List should have 25 free We have been using these rules of thumb for quite some time now and they work very well on the 8350 - if anyone has experience with similar numbers on a 6410 with 128 meg I would like to see your comments, as I plan to use similar numbers when we get our new machine. Disclaimer: While we use these values with great success, they may not work on all machines so use with caution. Make CERTAIN that you have a back-out plan when changing SYSGEN parameters, as it is a major bummer when your system refuses to boot back up after you have made a few changes!!! ______________________________________________________________________ / Rockford Lhotka INCSTAR Corp \ | Systems Administrator PO Box 285 | | incstar!lhotka@uunet.uu.net 1990 Industrial Blvd | \ 612/779-1701 Stillwater, MN 55082 / ----------------------------------------------------------------------