Distributed File Systems

[David Lang]

the EMC replication can be done over a DS3 link, it doesn't require
dedicated fiber.

it is however, very expensive.

Also like all disk sharing setups it has the problem that it is
replicating the raw drive without coordination with the mail software.

This means that unless you are useing a journaling filesystem you can end
up with a filesystem that isn't consistant (needs a fsck) so the failover
time needs to take this into account.

David Lang

On Sun, 20 Oct 2002, Jeremy Rumpf wrote:

> Date: Sun, 20 Oct 2002 17:19:13 -0400
> From: Jeremy Rumpf <jrumpf at heavyload.net>
> To: David Chait <davidc at bonair.stanford.edu>, info-cyrus at lists.andrew.cmu.edu
> Subject: Re: Distributed File Systems
>
> On Sunday 20 October 2002 04:07 pm, David Chait wrote:
> > Jeremy,
> >     While that would resolve the front end problem, in the end your mail
> > partition would still be a single point of failure. I'm trying to find a
> > way to do a real time replication of the mail partition between both
> > machines to allow for a complete failover.
> >
>
> I almost forgot, one way to perhaps accomplish this using a variant of what I
> posted before is to use EMC disk arrays and their ability to replicate over
> single or multi mode fiber at long distances. The failover/remount procedure
> of the second "buddy" node would mount the replicated EMC array instead of
> the local shared disk subsystem.
>
> I discounted this for two reasons:
>
> 1> Not everybuddy has two data facilities with dedicated fiber running between
> them (I happen to actually work in a place that has this, but it's not
> applicable to most).
> 2> Cost, Cost, Cost, and ummmmm Cost. Those EMCs are worth a nice house :).
>
>
> Jeremy
>