Don't let the disk pool fill up.
We see that normally migration will clean data off a disk pool a lot faster
than node backups put it on the disk pool.   So can you not set
highmigration=50, lowmigration=20 andthus make sure your disk pool never fills
up?  Migration will grab a tape drive every now and then when it drains the
disk pool to tape, but perhaps not for too long.

You might also consider using admin schedules to tweak the highmig and lowmig
parameters on the disk pool, to delay / encourage migration at different times
of the day.






ADSM-L AT VM.MARIST DOT EDU on 04/07/98 10:29:20 AM
Please respond to ADSM-L AT VM.MARIST DOT EDU
To: ADSM-L AT VM.MARIST DOT EDU
cc:
Subject: Large, slow backups/archives


I have several systems that have test oracle databases I need to archive
with great regularity. Because of the network interface, these systems
cannot drive my tape units at anywhere near efficient speed. As a
result, I run them into a disk pool and let this migrate to tape as it
fills.

How do I prevent these systems from going direct to tape if the disk
pool is full? These are test environments, and low priority - and the
original system is about 22 GB running through a 3.4 GB ADSM pool, so it
will fill from time to time. I *cannot* afford to have a tape unit tied
up for the length of time these systems will take to move an object to
tape.

My current 'technique' is to cancel the session when it goes to tape,
and let the retry process do recovery - but this is too labor-intensive.

Suggestions?

TIA

Tom Kauffman
NIBCO, Inc

kauffmant AT nibco DOT com