No, you are not overreacting. They are gamboling with backups.
Jim
"A Unix Admin" <sombody_at_somewhere.rr.com> wrote in message
news:Ig8w7.290620$aZ.58066199_at_typhoon.tampabay.rr.com...
> Thanks for your input Sean, that is what I thought. My other question is
> whether the backups that we do are reliable and/or recommended. I would
> think that if a transaction is being written at the same time as a split
> would be dangerous. Am I overreacting?
>
> A Unix Admin
>
>
> "Sean Fitzgerald" <sfitzgerald_at_centurytel.net> wrote in message
> news:E88w7.9389$tE2.4668714_at_feed.centurytel.net...
> > An easy way to use "split mirorrs" is
> >
> > 1. synchronize the mirrors
> > 2. issue "alter tablespace xxx begin backup" for each tablespace
> > 3. split the mirors
> > 4. issue "alter tablespace xxx end backup" for each tablespace
> >
> > Your database is in backup mode for only a few minutes at most and thus
> the
> > bad effects (generating blocks of redo) are minimized. I don't know why
> > anyone would go through the pain of a recovery every day just for a
> backup.
> >
> > Incidentally, if you use omniback with RMAN and split mirrors you will
do
> > exactly the above steps. Once the split mirrors are mounted on a backup
> > server, the database is put into mount mode and rman is used to backup
the
> > files. Oracle has several white papers on this and there are already
> > pre-built omniback utiltity scripts to achieve all of this. I am
> currently
> > using this method on a 3-node (all active) HP V2600 cluster running
Oracle
> > 8.1.6 (OPS). We have the cluster using an XP512 disk array with the
> > business continuance volumes. The split mirrors are mounted to an L2000
> for
> > backups (using RMAN and omniback).
> >
> >
> >
>
>
Received on Sun Oct 07 2001 - 22:14:11 CDT
