I see this type of question come up all the time with the very very
large database servers.

I don't see it as to what type of software you use
(Legato,ADSM,Arcserve,other...)

You are still limited by your network/tape speeds.

200Gb worth of databases on each server?  And you need to do a full
every day because the full
restore would take too long if you use a Backup Agent or an incremental
every day.

Basically for these large systems my opinion is; do not worry about the
restore time.  Make sure
that the database server is fully mirrored/raided so there would be no
need to restore all the
databases or the entire server in most cases of hardware failure.

That leaves the chance that if you lost the whole server due to a major
disaster (earthquake/fire/etc)
it would take a real long time to restore.  But remember you will have
to replace the hardware first
(how long would that take?). Calculate *that* into how fast you need to
do a restore.  Also consider
that you most likely will have the other 3 200Gb servers in the same
room that was damaged by
the earthquake/fire/etc...

If you have the hardware already available to recover onto then you
should probably be doing some
sort of database replication so you wont even have to restore it from
your backup server.

Use the backup server to restore files and things that were lost due to
people deleting them or
people wanting an older version back.

Have other plans in place for very large important servers that need to
be restored quicker
than the average tape drives/network can.  But of course always remember
that if all else fails
you do have a copy of that data on your backup server.

This is all my opinion of course, if anyone sees any flaws in my logic
please let me know I am just
learning ADSM.

--
Matthew Glanville
Matthew Glanville
mwg AT kodak DOT com
(716) 477-9371