> [...] I'm in no doubt whatsover of the importance of a
> valid catalog backup. The reason for the issue is simple -
> there are 50-100+ jobs running at any time of day. There
> is never any inactivity for long enough to get the 80
> gigs-worth of catalog data  backed up via the
> normal method.

Common, but not growing your backup resources to keep up with increased
backup requirments is not, IMO, a valid reason to accept increased risk
of not being able to recover in a disaster-recovery scenario.

> I had a long chat earlier with a ex-Veritas pal about this 
> and he came up with a plan as a workaround. (He called it
> the Devon Method, as that's where all the best fudge is made ;-) )
> 
> Firstly, yes, backup to disk.  That's easy, the database

Not recommended; see the SAG.  Sometimes it's easy to confuse
"completing the catalog backup requirement" with what will be needed
when your datacenter is toast and you (or your replacement) is trying to
rebuild at the DR site.

> volumes are
> already on SAN disk, so backup to disk is a trivial effort 
> for us here.
> Then reconfigure the catalog backups as follows:
> 
> master:/usr/openv/netbackup/db/images/master
> master:/usr/openv/var
> master:/usr/openv/volmgr/database
> master:/usr/openv/netbackup/db/[A-Zabd-hj-z]*
> 
> This will exclude the images completely. Then configure two standard
> policies to backup up the image directories ONLY as a daily 
> FULL with a two
> day retention alternating between the two. A copy could be 
> taken to tape
> and offsited from those.
> 
> How about that?

Your ex-Veritas pal has RTFM.  You should, too.  

The method you show is fully documented in the NetBackup Administrator's
Guide, both implementing the multi-tape catalog backup and recovering
using it.  It is designed for servers whose catalog will not fit on one
tape.  It still requires a correct catalog backup, complete with the
requirement that no other catalog-modifying activities, like backups,
are in progress.

The other documented method is catalog archiving, which also requires
NetBackup quiescence.  

Pre-NetBackup6 (online catalog backups), if you do not have quiescent
time for catalog backups, you have configured NetBackup incorrectly.
Add resources or change scheduling to allow a correct catalog backup, or
document that you and your management have made a business decision to
substitute disaster-recovery methods not recommended by the backup
vendor.