Jon and Stan thanks for your feedback!

2006/3/30, Jon LaBadie <jon AT jgcomp DOT com>:
> > Do you really need teh complexity of doing both virtual tapes, and
> > physical tapes? The only real advantage is in speed of recovery,
> > and, if you work at it hard enough, being able to recover, while a
> > backup is running.

I would agree.  Recovery has been a rare event here.  About one a month or less.

> > Sounds like your hardware is better suited to a "classic" Amanda
> > set up to me.

That is fine with me.  I like simple  :-)


> I had a similar thought.  You say your available hard disk space
> is not sufficient to hold a full set of level 0 dumps.  I think
> you said that was 600GB and you had only 400GB of disk.

Yes, approximately.  At any rate:   'full b/u > holding disk'


> In that situation, even if you spread out the dumps with incrementals
> and full dumps you could not hold a single dumpcycle's worth of
> dumps on the hard disk.  Assuming you had a 7 day dumpcycle, you
> might be able to save 3 - 4 days worth of dumps on disk.  And many
> incrementals would be of less value because their level 0 starting
> points were already deleted.
>
> I suspect you would be better off using that 400GB disk for holding
> disk space and saving your dumps to tape.
>
> What capacity is your tape?

Nominally 400G, but according to 'df -h' = 367G


>  I would suggest that when you start your
> production configuration you ease into the disklist entries (DLEs).
> If you have several hosts, each with several DLEs, enter all the DLEs
> into your disklist file, but comment them out with "#"s.  Then before
> each days' run of amdump, uncomment one or two DLEs from each host.

Good strategy.  We have about 40 computers each with 1-2 DLEs and a
file server with 14 DLEs.

Thanks and keep the advise coming, I need it  :-)


--
Enrico Indiogine
Parasol Laboratory
Texas A&M University

enricoi AT cs.tamu DOT edu
hindiogine AT gmail DOT com
979-845-3937