If you want to understand how TSM reads/writes you have to discuss it with
yourself - how do want this to happen:
See MIRRORREAD DB, MIRRORWRITE DB, MIRRORREAD LOG and MIRRORWRITE LOG
options in dsmserv.opt. Use them if you plan to use disks directly
controlled by TSM. Or you can create RAID-1 or use LVM mirroring ... Then
RAID settings or logical volume Mirror Write Consistency will control the
behavior. Be aware that MWC relies on jfs for synchronisation after
failure. For details you can look at "System Management Concepts: Operating
System and Devices".
I hope this helps a bit.

Zlatko Krastev
IT Consultant





Gerald Wichmann <gwichmann AT SANSIA DOT COM> on 11.10.2001 21:50:43
Please respond to "ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>
To:     ADSM-L AT VM.MARIST DOT EDU
cc:

Subject:        db/log allocation

I have a system that will have 2 36GB SSA disks and 2 9.1 SSA disks for
use to store my TSM DB/log. Now in the interest of performance, whats
the best way to carve this up? Obviously I will employ TSM mirroring so
that leaves me one of each drive. Would it be better to create two
RAID-0 arrays and store my DB/log on one, mirroring it to the other in
the interest of performance so that the TSM server reads/writes the data
to both drives in parallel (striping)? Or would it be better to create a
filesystem on one drive, and another filesystem on the other, then
create my db/log volumes on both in some way so that TSM reads/writes to
both drives at the same time?

I guess what I'm asking is I'm not entirely clear on how TSM
reads/writes to the db/log volumes when it's doing backups/restores and
I need to understand this in order to understand how to best carve up
these disks.

Thanks for the help :-)

Gerald Wichmann
System Engineer
StorageLink
408-844-8893 (v)
408-844-9801 (f)