We have experienced this phenomenon as well at 5.3.5.2 on a Solaris TSM Server.
The system has 32GB memory, 6 CPU. We used 20GB for bufferpool and verified that there was no paging going on. Performance was terrible! As a measure, expiration took about a week. All other processes, sessions, queries were similarly bad. Reduced buffpool to about 1/2 GB and performance is now great. Expiration now runs in about 3 hours. Other measures improved dramatically as well. Current DB stats: Available Space (MB): 204,800 Assigned Capacity (MB): 202,860 Maximum Extension (MB): 1,940 Maximum Reduction (MB): 75,212 Page Size (bytes): 4,096 Total Usable Pages: 51,932,160 Used Pages: 32,695,965 Pct Util: 63.0 Max. Pct Util: 63.0 Physical Volumes: 16 Buffer Pool Pages: 131,072 Total Buffer Requests: 306,981,255 Cache Hit Pct.: 99.72 Cache Wait Pct.: 0.00 Robert R. Price TSM Administrator Computer Sciences Corporation Phone: 412-374-3247 Fax: 412-374-6371 [EMAIL PROTECTED] Computer Sciences Corporation Registered Office: 3170 Fairview Park Drive, Falls Church, Virginia 22042, USA Registered in Nevada, USA No: C-489-59 ------------------------------------------------------------------------------------------------------------------------------------------------------------------------- This is a PRIVATE message. If you are not the intended recipient, please delete without copying and kindly advise us by e-mail of the mistake in delivery. NOTE: Regardless of content, this e-mail shall not operate to bind CSC to any order or other contract unless pursuant to explicit written agreement or government initiative expressly permitting the use of e-mail for such purpose. ------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Matthew Glanville <matthew.glanvill [EMAIL PROTECTED]> To Sent by: "ADSM: ADSM-L@VM.MARIST.EDU Dist Stor cc Manager" <[EMAIL PROTECTED] Subject .EDU> Re: Improving TSM performance - memory related setting 04/07/2008 09:43 AM Please respond to "ADSM: Dist Stor Manager" <[EMAIL PROTECTED] .EDU> Just a warning, don't necessarily go to 1/2 or 1/8th of your total physical memory.. If your server has 64 GB of memory, 8 GB (1/8th) for BUFPOOLSIZE is probably too high. I would keep it below 1 GB unless you prove to yourself with some testing that it is helping speed up the backups or restores. Just don't look at database cache hit rate. I think there's some inefficiency in how TSM database cache works, as it may have to search through that cached memory to find a hit. When you are caching 8 GB, that takes some time to search through all that. You'll see this as the dsmserv processes CPU use will increase the larger the BUFPOOLSIZE is. If it takes longer to search through that memory than it does to read the page from disk, you are hurting performance not helping it by caching in memory. Maybe DB2 TSM in version 6 will help this issue. I tried to put a problem in for it a few years back, but couldn't get beyond level 2 since no one understood what I was talking about, they kept thinking the server was slow due to the operating system using it's own page file to make up for the large memory setting, even though the 64 bit server had 32 GB of physical memory... Ahh well. I can use that other 30 GB for something else. or is it fixed now? Matt G. "ADSM: Dist Stor Manager" <ADSM-L@VM.MARIST.EDU> wrote on 04/04/2008 03:40:56 AM: > Bufpoolsize should be set between 1/2 and 1/8 of the among of the > total physical memory, if I remember well, that is, what best > practices recommend. > Also if you modify this parameter you will have to take a look at > the percentage hit cache (q db f=d) that can be increase (modifying > the bufpolsize), or perhaps decrease if the setting is incorrect. > Percentage hit cache must be above 99 %. > > I will also recommend to take a look at the performance tuning > guide, that will let you know more about many settings that can > increase the performance of your TSM server. > > Regards, > Bernaldo. > > > > ----- Mensaje original ---- > De: Paul Dudley <[EMAIL PROTECTED]> > Para: ADSM-L@VM.MARIST.EDU > Enviado: viernes, 4 de abril, 2008 5:44:02 > Asunto: Re: [ADSM-L] Improving TSM performance - memory related setting > > Currently the BUFPOOLSIZE is set to: > > BufPoolSize: 122,880 K > > The server has 3 Gb of memory. Is this something I can increase? > > Regards > Paul Dudley > > Senior IT Systems Administrator > ANL IT Operations Dept. > ANL Container Line > [EMAIL PROTECTED] > 03-9257-0603 > http://www.anl.com.au > > > > > -----Original Message----- > > From: ADSM: Dist Stor Manager [mailto:[EMAIL PROTECTED] On > > Behalf Of Skylar Thompson > > Sent: Friday, 4 April 2008 1:05 PM > > To: ADSM-L@VM.MARIST.EDU > > Subject: Re: [ADSM-L] Improving TSM performance - memory related > setting > > > > Paul Dudley wrote: > > > If I recall correctly there is a TSM system setting that relates to > the > > > memory available on the server, and you can check and adjust this > > > setting to improve performance of your TSM server. > > > > > > > > > > > > Can anyone remind me which setting it is? > > > > > > > > > > Are you thinking of BUFPOOLSIZE and LOGPOOLSIZE? The former controls > > the > > size of the database buffer pool (bounded only by the physical memory > of > > the machine), and the latter the size of the temporary transaction > space > > in the recovery log (up to 8192 pages IIRC). > > > > -- > > -- Skylar Thompson ([EMAIL PROTECTED]) > > -- Genome Sciences Department, System Administrator > > -- Foege Building S048, (206)-685-7354 > > -- University of Washington School of Medicine