ADSM-L

Re: DB backup -- a single point of failure??

1999-02-13 14:14:28
Subject: Re: DB backup -- a single point of failure??
From: Ray Schafer <schafer AT TKG DOT COM>
Date: Sat, 13 Feb 1999 14:14:28 -0500
"Agnew, Doug" wrote:

> As I have been building a disaster recovery plan with ADSM, I have been
> trying to identify and eliminate all single points of failure.  Now I've run
> up against one that won't go away, the database backup.
>
> As I read the manuals, I note that there is no way to create a duplicate
> copy of the database backup tape.  So, if I lose (through media failure or
> whatever) the latest database backup or ANY in the series of the latest full
> and subsequent incremental backups, I cannot restore my database to 'last
> backup'.
>
> Granted, if I have two servers and do a server-to-server backup of the
> database and then copy that archive file, I do get a copy of the database
> backup.  But dual servers is not in the current configuration YET and I
> can't find any other way to avoid this single point of failure situation.
>
> Am I just not thinking clearly or is this a legitimate problem??

If your database isn't too big, this solution will work:

We allocated a couple SSA drives and created a mirrored logical volume, then a
filesystem on top of that, then devclass of type 'FILE',  storing the daily
complete backup on the mirrored volume.  We keep only the 3 latest backups (run
delete volhist.... daily).  If we loose a disk drive, we can recover the
database from the mirror.

--
Ray Schafer                   | schafer AT tkg DOT com
Ray Schafer                   | schafer AT tkg DOT com
The Kernel Group              | Network Computing Consulting