Re: UNIX root backup/SAN disk image vs. Oracle Hot Backups

From: Mark Bole <makbo_at_pacbell.net>
Date: Thu, 02 Sep 2004 02:18:41 GMT
Message-ID: <5qvZc.14343$7a.283@newssvr27.news.prodigy.com>

Joel Garry wrote:

> idogan_tech_at_yahoo.com (Ibrahim DOGAN) wrote in message news:<6bf58828.0409010559.5ccd70bb_at_posting.google.com>...
>

>>Can UNIX root backup or SAN disk image backup be reliable as Oracle
>>database backup?
>>
>>I don't know internals about root backup or disk images but it sounds
>>like you may end up with inconsistent control file/datafiles when you
>>load them back ?
>>
>>is there anybody out there who knows how root backup/disk image works
>>and can tell whether they can replace Oracle hot backups?
>>
>>Thanks a bunch
>>i.d.

>
>
> As Sybrand sort-of intimated, they must work together. So, it depends
> how fast your mirroring works. The general idea is to
> [triple-]mirror, put all tablespaces into backup mode, break [extra]
> mirror, take out of backup mode, and copy the image at your leisure.
>
> Backup mode causes much more writing to logs (look up "fractured
> blocks" in the Oracle docs), so that's why RMAN is better versus
> copying one tablespace at a time over a period of time, it has a much
> lower impact on log writing. Image copies moot that, if they are fast
> enough. Some configurations take a long time to resilver the mirror,
> so, it depends, on transaction load, hardware, requirements, data
> size, etc.
>
> Of course, RMAN has a lot of other benefits that might outweigh a
> naive site-specific mirroring strategy.
>
> jg
> --
> @home.com is bogus.
> Well, shiver me slartibartfast!
> http://news.bbc.co.uk/1/hi/entertainment/tv_and_radio/3615046.stm

I've been moderately confused by this issue (not a good thing to admit, I guess). Try searching for the following article title from Sys Admin Magazine .. (I have the paper version, don't think the on-line version is available for free):

"Oracle Database Backups via UFS Snapshots and Enterprise Backup Products • Rick Moakley. Moakley describes how he uses UFS snapshots (and a little multi-tasking) to save significant time in the backup of an Oracle database. "

The OS-level journalling and snapshot facilities seem to be doing the same thing as redo logs in Oracle. Instead of worrying about updating the datafiles right away, the most important thing is to update the journal (redo). From this, you can re-create a read-consistent version of the file system at a given point in time. No need to wait for mirrors to be synchronized (or any mirrors at all).

If you are using hot backup or cold backup, it seems to me this should work fine and be as "reliable as Oracle".

All backups are site-specific, if not naive ;-)

--Mark Bole

(Never read the Hitchiker's Guide... my loss... so far...) Received on Wed Sep 01 2004 - 21:18:41 CDT