Willem, what I can see from the accounting log is that the backup has an enormous amount of media wait. There are two records for that backup which I interprete as the two mount points. One mount point should be for the tape that has the meta data and the other mount point for the data itself. Under normal conditions the first record has as many seconds as the backup duration while the other record has 0 seconds. When I experience the delay the second record has as many seconds accounted for as the first record.
Of course what I don't understand is why on the same client on the same day I get different responses for two databases. Any thoughts are welcome. Yiannakis PS. On Dave's question : I have created a disk stg pool that then migrates to tape. For small size databases it goes to disk first. Larger databases go directly to tape because the disk pool is small. -----Original Message----- From: ADSM: Dist Stor Manager [mailto:[EMAIL PROTECTED] Behalf Of Willem Roos Sent: 09 December 2003 12:55 To: [EMAIL PROTECTED] Subject: Re: SQL TDP configurations [servername:] set accounting on q status will tell you whether its on, i'm not sure whether its on by default. The file is in /usr/tivoli/tsm/server/bin or Program Files\tivoli\tsm\<servername>. It is described in the administrators guide and on http://people.bu.edu/rbs/ADSM.QuickFacts, search for dsmaccnt.log. I have an Excel spread sheet which sort-of interprets such a file if you want it. -----Original Message----- From: ADSM: Dist Stor Manager [mailto:[EMAIL PROTECTED] On Behalf Of Yiannakis Vakis Sent: 09 December 2003 12:37 To: [EMAIL PROTECTED] Subject: Re: SQL TDP configurations Willem, how can you set accounting on for TDP SQL client ? I searched but couldn't find anything. Yiannakis -----Original Message----- From: ADSM: Dist Stor Manager [mailto:[EMAIL PROTECTED] Behalf Of Willem Roos Sent: 09 December 2003 11:16 To: [EMAIL PROTECTED] Subject: Re: SQL TDP configurations Are they scheduled from the same node - maybe the one backup is waiting for the other scheduled one to complete? Is it lan-free? Also, check that accounting is switched on and check the dsmaccnt.log file - it will give you session accounting with lots of interesting wait times (media wait, comms wait, idle wait), maybe the one backup suffers lots of media wait? -----Original Message----- From: ADSM: Dist Stor Manager [mailto:[EMAIL PROTECTED] On Behalf Of Yiannakis Vakis Sent: 09 December 2003 07:46 To: [EMAIL PROTECTED] Subject: Re: SQL TDP configurations Hi, I have the same situation here for no apparent reason. I do weekly backups using TDP SQL. Three big databases on the same client behave differently on the same day. The MIS db is 180GB and takes about 3hrs. The TRN db is 120GB and takes 2hrs to complete. The BAL db is 80GB and take 1.30hrs. This happens for two weeks. Then for no apparent reason the TRN db requires 5hrs to complete while the MIS db still takes 3hrs and the BAL db 1.30hrs. So the TRN db backup just slows down for two or three weeks. Then with no change it gets back to normal. I have checked network connections, routing configuration, database reorganisation - all is fine. Any input would be welcome. Yiannakis -----Original Message----- From: ADSM: Dist Stor Manager [mailto:[EMAIL PROTECTED] Behalf Of David Benigni Sent: 09 December 2003 04:58 To: [EMAIL PROTECTED] Subject: SQL TDP configurations Simple question. What is the optimal why to back up MS SQL databases? Whats the ideal pool configuration etc? Right now I'm running into trouble with the way Tivoli was originally figured for our systems. The SQL tdp's sit on the client and back up to a tape pool just for sql backups. The tape pool is backed up to our drm pool then. The problem is our one sql database takes a long time to backup, and it pushing the schedules all off. The other weird thing is, the sql tdp for no apparent reason decided to cut its throughput in half. Needless to say the backups take twice as long. And the machine itself has not changed. Any thoughts would be great. TIA. David Benigni
