Hello Mark
When the restores were running on the master/media/client
systems, did they both have the same level of usage outside of the restores.
That is was the cpu busy, i/o usage pattern, network usage etc at the same
levels for both restores.
If the user backup was running when the systems was under heavier
usage and the master scheduled backup was running when the systems was lightly used,
this might explain your differences.
I am not sure how much different the process flow for a user
backup vs a backup scheduled from the master. But it might be the relative
system priority of the inet.d started processes vs the process's started by the
user backup.
len
From:
veritas-bu-bounces AT mailman.eng.auburn DOT edu
[mailto:veritas-bu-bounces AT mailman.eng.auburn DOT edu] On Behalf Of mark.a.goodchild AT bt DOT com
Sent: Wednesday, July 04, 2007 10:03 AM
To: veritas-bu AT mailman.eng.auburn DOT edu
Subject: [Veritas-bu] Performance issues with user initiated backups.
Hi all
Environment
:
Master
Netbackup 6.0 MP4
Solaris 10
Media server
6.0 MP4
HP-UX 11
IBM 3584
Library San Attached
IBM LTO2 Drives
Media server
is the oracle client too so it’s a local backup.
Data :
/u21/oradata/D1
118gb 169 files
User Backup
of above writes to drive1, tape1 at 13mb/sec
Scheduled/immediate
backup writes to drive1, tape1 at 27mb/sec
We believe
that there is so much communication going on between all the netbackup
processes regarding files and updates when a user backup is run.
Can anyone
think of anything else that would slow this backup down to half the speed.
We have done
numerous tests on different drives, tapes and even chosen different data
locations, and come up with the same results.
Cheers