Hi Doug,
I agree with Wanda entirely. This very debate arose within out organization
when we deployed TSM a couple of years ago. We have about a dozen remote
offices which have no TSM server at their local site. We implemented our
configuration almost precisely as Wanda described, minus the sub-file
backups. Our business requirements are such that we have one day to restore
a down server in these remote offices. Thus, we have installed DDS4 tape
drives in each remote server as well as one on our TSM servers. Should
failure occur, we generate a backupset (about 2-3 hours), travel to the
remote office (1-6 hours), repair the server(?? hours) and then restore from
the backupset 2-3 hours). This solution has worked well for us as we have
used it to recover several servers, and was cheap to implement! Hope that
helps!
Chris Murphy
IT Network Analyst
Idaho Dept. of Lands
Office: (208) 334-0293
cmurphy AT idl.state.id DOT us
-----Original Message-----
From: Prather, Wanda [mailto:Wanda.Prather AT JHUAPL DOT EDU]
Sent: Tuesday, October 21, 2003 9:11 AM
To: ADSM-L AT VM.MARIST DOT EDU
Subject: Re: TSM Downward Scaleability
Actually there are some 2-drive tiny LTO and AIT libraries that are very
affordable; that's the best alternative.
If you can't do that, consider backing up over the WAN.
Whether it's practical depends on the type & size of the remote clients, but
if they aren't huge, it's possible.
The initial backup will take a long time (so what, do it on the weekend).
But If you use client compression AND enable subfile backup, the daily
backups may be very manageable.
That will give your remote sites the ability to do ad hoc restores, usually,
with no problem.
And your site becomes their "vault", so disaster recovery is taken care of.
That leaves the big problem to be LARGE restores - e.g., suppose a hard disk
dies on the remote client and requires 30GB to be restored.
If you are lucky enough to have a fairly homogenous client population, for
this case you could consider keeping a spare server at your main site;
rebuild it locally and then fed-ex it out to the remote site instead of
taking days to restore with limited bandwidth.
If they need rebuilds faster than that - they gotta get budgets for local
tape!
-----Original Message-----
From: Douglas Currell [mailto:dlcurrell AT YAHOO DOT COM]
Sent: Tuesday, October 21, 2003 10:54 AM
To: ADSM-L AT VM.MARIST DOT EDU
Subject: TSM Downward Scaleability
The organization that I work for deploys TSM quite
sucessfully at its large main sites that serve some
+4000 nodes. It is very apparent that TSM scales
upwardly very well but I believe that scaling down is
something else. MY question is this:How can similar
services be delivered to sites where there are less
than ten nodes, limited bandwidth, no system
administrators and, most importantly, tiny budgets.
The organization can comfortably absorb the price of
software and TSM licensing for these sites but
there';s no budget to equip each site with a dual tape
drive library. What are the alternatives? An
autoloader, for example.??
______________________________________________________________________
Post your free ad now! http://personals.yahoo.ca
|
