ADSM-L

Re: Your advice wanted!

2002-06-27 17:00:12
Subject: Re: Your advice wanted!
From: Zlatko Krastev <acit AT ATTGLOBAL DOT NET>
Date: Thu, 27 Jun 2002 23:52:56 +0300
Have in mind that due to lack of random access copypools you have to plan
also reclamation there (of course if copypools is used at all). I would
prefer to mirror primary diskpool volumes (of course DB&Log too) thus
getting random access copies and still protected against HDD failure.
Do not forget to schedule backups to file devclass copypools are using.
Backup of volhistory&devconfig would help but even without them files from
DB backups have .DBB extension and are easily recognizable. And they may
go off-site if using Alex's idea for disks exchange (there is no
hot-swapping for IDE but we know the cages for quick "cold"-swapping).
Alex, you will not be able to send Shark's disks off-site because ESS LIC
will complain but try to send the whole Shark off-site :)

Zlatko Krastev
IT Consultant




Please respond to "ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>
Sent by:        "ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>
To:     ADSM-L AT VM.MARIST DOT EDU
cc:

Subject:        Re: Your advice wanted!

Hi.

I don't see much return on making those primary disk storagepools
sequential, because once they get tape hardware, you can just "move data"
or
migrate the backup data off of the random diskpool volumes.  In fact, it's
more of a headache because you'll have to start reclaiming them and
whatnot.
Definitely stay with random access disk volumes in your primary diskpools.

For copypool, I wonder.  Since your installation is so small, I wonder if
you can get some hotswap drive bays (do those exist for IDE?), buy 2 more
IDE hard drives (they're fairly cheap, aren't they?), and start an offsite
rotation of your copypool disks.  That would be cute.  And much cheaper
than
investing in a new tape infrastructure to begin with.  Hmm... I wonder if
I
can do that with Shark disk.  But your copypool would have to be
sequential,
so that would complicate matters.

If you can figure out how to do sequential volumes and reclamation and
whatnot on disk, I would use those two disks as copypool, with or without
the extra 2 disks for offsite.  Then if you have an application based
corruption of your primary diskpool volumes, your copypool has a good
chance
of surviving that because it's more of an asynchronous "mirror" process.
Synchronous mirroring would be more vulnerable to application based
corruptions.

Have you given any thought to how you're going to manage your dbbackups?
It's a good thing to have them on some other machine or media.  You could
back up your database to disk and ftp it to another machine, or mount
remote
disk and back up to it, or half a dozen other variations.

Good luck.

Alex Paschal
Storage Administrator
Freightliner, LLC
(503) 745-6850 phone/vmail

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