Yes the disk pool being close to full can do it.  TSM will try to
pre-allocate the storage on the disk pool for the file being backed up.
If there isn't space for it, it will pass the backup to the seconday
pool, your tape drives). If I remember this correctly, once the backup
starts going direct to tape it seems to keep doing that.  
        There is also a max file size on the diskpool.  If the file, (if
it is through an API it treats the backup as 1 file per SESSION),  is
larger than this value it goes straight to tape.
        If you really want to speed things up, increase your MAX MOUNT
POINTS on your nodes definition.
Matt

-----Original Message-----
From: ADSM: Dist Stor Manager [mailto:ADSM-L AT VM.MARIST DOT EDU] On Behalf Of
PAC Brion Arnaud
Sent: Monday, April 05, 2004 8:25 AM
To: ADSM-L AT VM.MARIST DOT EDU
Subject: Multithreaded session writes on tape although primary pool is
on disk ...

Hi all,
I have an AIX client (5.1.6.9, TSM server at 5.2.2.1), which needs to
backup it's files as fast as possible, therefore I modified it's dsm.sys
to give it "Resouceutilization 10". On TSM server this node has "max
mount points  2", and belongs a domain where primary pool (called
diskpool_aix) is on disk. This diskpool has a next storage pool called
"tapelto1_aix" which is collocated, and has no max size threshold.
Since I increased this resourceutilization parameter, it looks like the
client regularly sends it's backups directly to tapelto1_aix, even
initiating two parallel tape mounts, without going thru diskpool ...

Activity log looks like this (node name hidden for security reasons):
Date/Time Message --------------------
----------------------------------------------------------
04/05/04 02:04:48 ANR0406I Session 419574 started for node XXXXX (AIX) (
Tcp/Ip XXXXX (50490)). (SESSION: 419 574)
04/05/04 02:19:28 ANR8337I LTO volume 000215 mounted in drive LTO1DR6
(/dev /rmt6). (SESSION: 419574) 04/05/04 02:19:43 ANR0511I Session
419574 opened output volume 000215. (SES SION: 419574) ...../....
04/05/04 02:04:52 ANR0406I Session 419575 started for node XXXXX (AIX) (
Tcp/Ip XXXXX (50491)). (SESSION: 419 575)
04/05/04 02:21:06 ANR8337I LTO volume 000217 mounted in drive LTO1DR4
(/dev /rmt4). (SESSION: 419575) 04/05/04 02:21:17 ANR0511I Session
419575 opened output volume 000217. (SES SION: 419575) Off course this
is not really what I intended to get : I now have a node monopolizing
several tapes with very low occupation, that can't even be reclaimed ...
Nothing special noticeable at the same time, except that a migration
happenend for this disk pool approximatively 20 minutes later, so it was
quiet full. Could that be a reson for such a behaviour ? Any other
explanations, ideas ?!
Thanks in advance !


Arnaud Brion

***********************************************************************
Panalpina Management Ltd., Basle, Switzerland, CIT Department
Viadukstrasse 42, P.O. Box 4002 Basel/CH
Phone:  +41 (61) 226 11 11, FAX: +41 (61) 226 17 01
Direct: +41 (61) 226 19 78
e-mail: arnaud.brion AT panalpina DOT com
***********************************************************************