ADSM-L

Re: Very Ssllloooooowwww Backups

1998-09-21 08:10:26
Subject: Re: Very Ssllloooooowwww Backups
From: Roger Hohmann <Roger_Hohmann AT WESTLB DOT DE>
Date: Mon, 21 Sep 1998 14:10:26 +0200
Hi Alfred,

If Your diskpool becomes full during the night, migration starts, but is a
low priority job. So it waits until it gets enough resources (example: If
Your tape drives, cpu cycles etc are limited, migration is the last who
gets one. Migration will work when all backup is finished. So Your disks
are full the whole night, all backup goes to tape. You need minimum 1 Drive
per concurrent session.)

If You don't know how busy Your diskpool is- look in Your activity log how
often migration
occurs. If it's once a day, You might empty them in the before Your backups
starts (or in the morning after backup) ("upd stgp backuppool hi=1 lo=0"
will initiate migration,
"upd stgp backuppool hi=90 lo=70" will change to normal operation after
migration ends.)
This will fit if you have not more than one migration per day.

If you have more than one migration per day, You have to tune your
threshold parameters, we are using something like lo=5,hi=40. Migration is
a low priority task, so You should start early to get the work done.

For detailed information You should use accounting. Then You can see how
many data is moved per night, which node etc.

Regards

Roger
<Prev in Thread] Current Thread [Next in Thread>