[ale] disk backup!

[Dan Newcombe]

On Wed, 18 Jul 2001, Joseph A. Knapka wrote:
> storage in terms of data density. (I know I'll probably eat those
> words next year, but whatever.) That being the case, and with
> hard-disk storage being dirt cheap these days, it probably makes
> sense to build RAID into every server, and just replace
> the disks as they fail. You could build a network server just
> for this, and have the other machines dump their data to it
> each night.

While it is fast and easy, it is not something I'd want to rely on.  I
would think that disk would be great for the last nights backup, replaced
everynight.  However, removeable media has that one advangtage - you can
get it out of there.  Take 7 DLT's and put them in a bank vault.  Switch
them out every night so you have other levels of backup to fall back onto.
While we have some sort of RAID on all our servers for redundancy, that
only protects against hardware faults and downtime, not human faults. 

No matter what you do, just make sure that your backups are working and
getting what they need on a regular basis - test them.  I've heard horror
stories from people that were sure the backups were working, because the
tape drive didn't give errors, but it just wasn't writing anything but
garbage to the tape.  Or then there are people who were backing up Oracle
databases while they were up and running, so when they had to recover, one
datafile had a different "version" number than the other - and therefore
an invalid (dead) database.

	-Dan

--
To unsubscribe: mail majordomo at ale.org with "unsubscribe ale" in message body.