ADSM-L

Re: Tape Copy pools and Colocation

2001-10-25 11:10:57
Subject: Re: Tape Copy pools and Colocation
From: Jack McKinney <jackmc-adsm AT LORENTZ DOT COM>
Date: Thu, 25 Oct 2001 10:08:16 -0500
    When you run BACKUP STGPOOL with the PREVIEW=VOL or PREVIEW=YES options,
then the tape that it reports are that ones from the primary storage pool
that it is going to need to read from, not the ones it will need to write.
    In other words, it is saying that it needs all 27 primary tapes, not
that it will need 27 output tapes.  It cannot estimate that since compression
can change the total stored space on each tape.
    The idea with preview is that if you backup your primary storage to this
storage pool again next week, only 6 tapes may have new data on them that
needs to be backed up, so you can find out which tapes you need

    The copy pool tapes should be written front to back, even though the
primary storage is collacated.  Collacation is not inherited, but you
can specify this in the copypool when you define it.  The only reason to
do this is for DR purposes.  If you lose your entire server room, but
your copy pool volumes (including a DB backup!!!) are offsite, you can
restore the database and then start restoring clients from the copypool
itself, without having to restore the copy pool to a primary pool.

Big Brother tells me that Kyle Sparger wrote:
> Hello,
> 
> I'm starting to implement tape copy pools.  My primary pool has colocation
> enabled, in order to speed up restores.  My primary storage pool currently
> is using 27 tapes at varying degrees of usage.  When I did a test BACKUP
> STGPOOL, the output indicated that it wanted to use another 27 tapes, even
> though that would end up leaving unused space on each tape.
> 
> I expect that the copy pool will not be used a whole lot;  I intend to
> move these tapes out of the library after each backup, and as such, prefer
> density and tape efficiency to speed of recovery from these tapes.
> 
> I was wondering if there is a way to have it use the copy pool such that
> it uses the fewest tapes possible, filling up each tape fully before
> pulling another one from scratch.
> 
> Or, does it do this already, and is the test run just being
> ultra-conservative?
> 
> Thanks,
> 
> Kyle Sparger - System Administrator
> ksparger AT dialtone DOT com - http://www.dialtone.com
> Voice - (954) 581-0097 x 122
> Why do programmers get Halloween and Christmas mixed up? OCT(31)=DEC(25)
<Prev in Thread] Current Thread [Next in Thread>