Re: ASM/RMAN sanity check

From: Niall Litchfield <niall.litchfield_at_gmail.com>
Date: Thu, 25 Apr 2013 06:59:48 +0100
Message-ID: <CABe10sbks5AXnn-E8vNOohrKKiyNW-4vmQHrYPP8ozcB-QYDaA_at_mail.gmail.com>

Some things to consider from my experience. 1) are you planning to put several databases on the same storage LUNS (diskgroups). If so can they always be down together? We have a data + fra per database precisely because we clone the disks to create non prod databases and only want 1 database owner to have to approve the downtime.

2) your upgrade scenario sounds like the classic use case for guaranteed restore points, not storage snaps We do this several times a week, it just works.

3) despite what I said in 1 above Oracle currently recommend just the 2 diskgroups (but then they also recommend using RMAN for backups) On Apr 25, 2013 3:44 AM, "Maureen English" <maureen.english_at_alaska.edu> wrote:

> We are moving databases to a new 11.2.0.3 Linux cluster, using RAC/ASM.
> The databases will be RAC One Node databases.
>
> In the past, we've always shutdown our databases every week and taken
> cold backups or snapshots of the database. We're trying to develop a
> backup strategy for the new server and I'm hoping that I can get some
> feedback from those with more experience to let me know if I'm on the
> right track or not.
>
> For our small databases, I'm planning to have the standard DATA and FRA
> diskgroups. The small databases will share space in these two diskgroups.
> RMAN backupsets will be created in the FRA diskgroup and then backed up
> to tape and deleted. This is what I've found to be "Oracle's Best
> Practice".
>
> For the large databases, we need the ability to do snapshots in addition
> to hot backups. My plan for these databases is to have DATA and LOG
> diskgroups for controlfiles, redologs, datafiles and archivelogs for each
> database. When we need to do a snapshot, we can shutdown the database,
> snap the 2 diskgroups and bring the database back up. I'm also planning
> to have an FRA diskgroup that can be used for the RMAN backupsets that
> can be written to tape and deleted.
>
> Does it make sense to have the 2 diskgroups per database and 1 shared FRA
> diskgroup for these large databases, or is there some other idea that makes
> more sense?
>
> One reason for the snapshots is because we will often take multiple
> snapshots
> during application upgrades so we have a point to quickly restore to if
> there's
> a problem with the upgrade. These need to be done when the database is
> down,
> and they need to be done quickly so the upgrade can continue in a timely
> manner.
>
> We're currently okay on being able to take a snapshot of an ASM diskgroup,
> but
> the backing up of it, and the recovering from it, are still in question.
> The
> system administrators are working on that....
>
> Any comments/suggestions?
>
> - Maureen
> --
> http://www.freelists.org/webpage/oracle-l
>
>
>

--
http://www.freelists.org/webpage/oracle-l

Received on Thu Apr 25 2013 - 07:59:48 CEST