From: Timothy J. M. <tm...@mo...> - 2008-01-18 03:24:35
|
Les Mikesell <le...@fu...> wrote on 01/17/2008 01:24:24 PM: > Timothy J. Massey wrote: > > > And there's no > > coordination between the two instances: the moment the copy is > > finished, they're two separate pools with separate schedules, etc. > > Which means, of course, that if something goes horribly wrong on one > (like the admin typing 'rm -rf ...' in the wrong place) you'll still > have the other copy at its last snapshot state. > > > Which is why, of course, you have to leave BackupPC shut down on the > > other box. DRDB helps: it eliminates both shutting down BackupPC and > > rsync limitations for duplicating the pool. > > DRDB might help propagate the disaster in real-time. I'm not convinced > that is desirable. You are correct; but that's what my second idea (replication of individual backups) is for: *automatic* and hands-off movement of backup data from one place to another. > > You could also do the same > > thing with breaking a RAID-1 mirror: Les brings this up every time! :) > > It works - and takes minimal down time. You could probably do this over > iscsi if you wanted to eliminate the physical disk swap. Sure, as far as it goes. I would like to have something that is managed completely under BackupPC. > > But you're still left with two indepenent servers, or at best an > > active/standby cluster. > > And in the backup business, that might be a good thing. When I > raid-mirror, I rotate the target disks so I never overwrite my last good > copy. I want two servers with two pools, but I do not want the load on my *clients* of having to back up data to two different servers. We have the same goals (redundant copies of the data), but different ways of doing it. > But I'm not sure it is so trivial to get the NFS performance you would > need from such a file server. I am *so* *not* talking NFS. Something more like GFS and iSCSI... :) Tim Massey |