1999-12-29-20:34:36 E Hunter:
> I'm looking for infomation on NetApps because the company I'm working
> for is considering a 760 for our network.
Congratulations! Netapps aren't the cheapest disk available, but they may just
be the nicest.
> I have been told by someone who is only vaguely familiar with the NetApps
> that the filesystem has to be maintained in some way by running a _wac_
> command, but I have been unable to find any information on such a command in
> a week of searching. Can anyone fill me in on the use of this _wac_ command,
> what it does, how often it needs to be done, and whehter or not the filer
> requires downtime to do it?
Lurking on a list like this could give you the impression that this is a
routine maintenance chore, but it ain't:-).
When you get at something like the toasters list, you are seeing a tight
population of a lot of the heaviest users of netapps, and they only tend to
post things when stuff blows up. And if you follow the analysis to its end,
it's surprising how often the final answer ends up being "boy, I shouldn't
have done that, next time I'll RTFM first".
I've only worked in two companies that used Netapps, and that only for maybe
two years altogether, but I ended up working with about a half-dozen of the
gizmos over time, and they were invariably plug-'em-in and they go, then you
get to forget them.
The trickiest maintenance chore isn't intrinsic to Netapps as such, but rather
comes with the territory whenever you start pushing up near a terabyte of
data; big horking disk farms exceed the capacity of any available tape media.
So backups are no longer a no-brainer. Brains required. Netapp does everything
to make it as easy as possible. I'd say WAFL's snapshots are the biggest
reason for not considering anything _other_ than a netapp for storing
terabytes. For a lot of purposes, my favourite backup strategy is to (a) have
a cold spare netapp in the rack (one such cold spare is plenty for many hot
toasters), (b) make sure you have a couple of times as much disk space total
as you need, so you can collect plenty of snapshots, and (c) occasionally spin
a stack of DLTs to back up a candidate snapshot, however often you need
archival coverage. For offsite coverage I'd still go with online replication
if I could; I've replicated a _lot_ of data with rsync-over-ssh, if you're not
in a hurry it just keeps on slogging. You do have to carve the job up into
subdirs and sync 'em separately, since rsync builds some O(nfiles) data
structures. I would use snapshots for "oops, I wish I hadn't deleted that"
coverage, I'd trust the toaster's raid to protect against disk loss, leaving
only archival tapes that have to be spun. And DLT juke boxes that can handle
22 tapes aren't that expensive, heavy, or bulky. I recently set up an ADIC
FastStor 22 <URL:http://www.adic.com/> and it's quite nice. That's somewhere
around 1.5TB worth without you having to get up and change tapes. As long as
you aren't in a hurry (_love_ them snapshots!) you don't have to work too
hard.
-Bennett
This archive was generated by hypermail 2b29 : Tue Jan 04 2000 - 10:20:56 EST