[OpenAFS] Replica sites

[Stephen Brown]

On Tue, 10 May 2005, ed wrote:

> I don't want a single system that could fail, and I don't want a huge
> hardware budget, even if it takes me 4 months to find a solution it's
> more worthwhile as I can apply it to other internal requirements where
> data is possibly brought down through a single failure.

What we do here is put e-mail in each users' disk volume -- each user
volume has a Mail, home, and pub directory in it, with appropriate ACLs.
User volumes are spread over many servers, and the multiple POP/IMAP and
SMTP servers mount AFS using the client.  Loss of one user volume server
takes a fraction of folks offline, but loss of one mail reader /
delivery server doesn't affect anyone (except the admins...) other than
a bit slower response.

Steve Brown
sbrown7@umbc.edu