In regard to: Re: [Imap-uw] Mix format downside?, Dan Pritts said (at...:
On Wed, Jan 17, 2007 at 06:21:11PM -0600, Tim Mooney wrote:
Believe it or not, it can take longer to do a true incremental of a
large filesystem (e.g. 250 GB in size, 210 GB used, 3.5 million files)
where almost nothing has changed than it takes to do a full backup of
the same filesystem. We've seen that several times on our largest
filesystems, which are generally Linux boxes with ext3 filesystems
(one of them is Linux with XFS).
We're using EMC/Legato NetWorker for our backups.
This may be obvious to you, but just in case:
it sounds like you are shoeshining your tape drives with your
incrementals.
In general, no, but once a backup group is down to just the largest straggler
filesystems, there is definitely some shoeshining going on.
To prevent that, two things come to mind:
- multiplexing backups from many hosts onto one tape
We're making heavy use of multiplexing already, but even with
multiplexing, you're eventually left with just a few filesystems that
are finishing up. It's not uncommon to see periods of several minutes
where none of the remaining clients have any data to send.
It is admittedly rare that an incremental takes longer than a full, but
we've seen it.
- staging your backups to disk (on the backup server) and then spooling
them to tape. Veritas netbackup has this feature, i presume legato
does too.
It does support it and we've experimented with it a bit, but we don't
have it in production yet. It's definitely something we're looking
forward to, for this reason and others.
Tim
--
Tim Mooney [EMAIL PROTECTED]
Information Technology Services (701) 231-1076 (Voice)
Room 242-J6, IACC Building (701) 231-8541 (Fax)
North Dakota State University, Fargo, ND 58105-5164
_______________________________________________
Imap-uw mailing list
Imap-uw@u.washington.edu
https://mailman1.u.washington.edu/mailman/listinfo/imap-uw
