This is exactly how we backup our more than 200 Oracle databases.  We
have a script (actually, a rather large suite of scripts) that create
the backup in a a staging area on the db server.  The backup is
completely contained under a directory that contains the db name,
date, time, type (hot/cold), etc.  TSM them come in and backs up the
staging area, as well as log files and other stuff.  Our use a
management class with 2 versions exists, 1 deleted, 1 retain extra,
and 90 retain only.  Basically, were not interested in anything but
the most recent version, and, when it's deleted we keep it around for
90 days.  Yes, if the backup fails for some reason there's a backup
in TSM that's bad.  The scripts do extensive logging so we can tell
how/if it worked.  In general, backups have so little problem I can't
remember the last time a Oracle hot failed.

Rick

On 1 Jun 2001, at 0:58, Suad Musovich wrote:

> We are about to start backing up our enterprise databases.
> They are comprised of about 15 Oracle instances ranging from 7GB to 25GB.
> About a third are being backed up nightly and the others are either 1 or 2 
> times
> weekly.
>
> On the server side, we will set up a seperate stgpool hierarchy as they there
> will be expectations for QOS etc. (I will seperate them out onto a 2nd server
> later this year)
>
> There are no TDPs available (Dynix) so we have to use the command line client
> to backup. The DBAs run the database in backup mode and cpio the files to
> a staging area, then the TSM client will scoop them up.
>
> They want a 30day retention period with one backup a month kept for a year 
> (acheived
> by a management class flag on dsmc)
>
> Question:
> Should we run incrementals or archive the files considering we have a date, as
> opposed to frequency, retention requirement?
>
> Suad
> --
>