Don't know anything about Quick I/O, but we have several Solaris TSM
servers that used to use filesystem volumes (on a VxFS filesystem) for
database, disk pool, and recovery log.  Upon implementing the
recommendation to switch to raw volumes (we did volumes under VxVM
control, but with no filesystems on them - just using
/dev/vx/rdsk/tsmdg/<volname> as the volumes inside TSM), and noticed a
*SUBSTANTIAL* performance increase - for instance expiration ran through
to completion for the first time in *YEARS*.  Not sure about the backup
performance though - it is fairly good, though.

Therefore, my recommendation from experience  is to use raw volumes, and
not any fancy Quick I/O garbage.

-Jon

-----Original Message-----
From: Forrest Payne [mailto:tsm_user AT YAHOO DOT COM] 
Sent: Friday, February 06, 2004 10:27 PM
To: ADSM-L AT VM.MARIST DOT EDU
Subject: Using Veritas Quick I/O on your TSM Server


We are running TSM V5.2.2.1 for Solaris on a Solaris 2.8 server. I am
not a Unix admin. The Unix support team loaded and licensed Veritas
Quick I/O. According to the TSM Performance tuning guide you either want
to use Quick I/O or RAW volumes when defining DB, LOG, or Disk pool
volumes on Solaris. Because our Oracle Administrators use Quick I/O we
decided to use it on our TSM server. I was under the assumption that all
you have to do is install Quick I/O and everything will be faster. Due
to the fact that we are currently getting extremely poor performance
when backing up to disk as opposed to tape I believe there must be more
we have to do in order to get the bennifits of Quick I/O.

Is anyone out there running TSM on a Solaris Server using Quick I/O? If
so what did you have to do in order to get Quick I/O to speed things up?



---------------------------------
Do you Yahoo!?
Yahoo! Finance: Get your refund fast by filing online