[OpenAFS] Replication on User Vol

Steve Devine sdevine@msu.edu
Tue, 24 Apr 2007 10:25:55 -0400

Melvin Wong wrote:
> Hi,
> I understand that replication is not suitable for volumes that change
> frequently like user volumes. Does this means that if I have the user
> volumes spread across many fileservers and if one of the fileserver
> dies, users whose data is in that fileserver will have no access to it
> until I recover from backup? Is there a recomended way to maintain
> large user volumes with any downtime? For example how does a
> university maintain these large user volumes for her students?
> tks,
> melvin
We have *lots* (150,000 plus) of user volumes spread out over 20+
dedicated fileservers. The way we keep the data up with low downtime is
by using mirrored high quality disk systems. We monitor the servers 24*7
and if we need to reboot or otherwise service a server we off load the
volumes to a spare machine before we do anything. In the end there is
always the possibility that catastrophe can strike but all you can do is
keep good backups and follow best practices when dealing with your servers.

Steve Devine
Network Storage and Printing
Academic Computing & Network Services
Michigan State University

506 Computer Center
East Lansing, MI 48824-1042

Baseball is ninety percent mental; the other half is physical.
- Yogi Berra