Coda File System

Re: Larger then 10GB Coda server?

From: Tim <jabbo_at_mindless.com>
Date: Mon, 10 May 1999 18:25:08 -0400
".Oliver_Thuns." wrote:
> 
> >> What's the difference between AFS and Coda?
> >
> >Was that a joke?
> 
> No

Ok, then please accept my humble apologies.


> >AFS is the predecessor to Coda.  (note the reference to "AFS2" on the
> >Coda pages)
> 
> And what are the advantages of coda?

They are eloquently explained on http://coda.cs.cmu.edu/ if you _really_
care.  If you don't you can rely on my pathetic synopsis: Coda is DFS
but it's Free and can "heal" disconnected clients with inconsistent
states, which requires keeping track of large amounts of metadata and
thus limits the current scalability of the system.  It is this
double-edged sword that one encounters when deploying Coda (for fun and
profit)... it allows something which is otherwise unavailable save for
the limited capabilities of a partially-replicated database client, but
being less mature than either the databases or filesystems it supplants,
Coda requires a bit of patience, bravery, and exploratory zeal.  And it
doesn't scale up yet.  Oh, and it has the administrative complexity of
AFS, which is not a trivial thing to learn.

Note that my employer doesn't see this as a problem ;-).  But my former
employer (CTG/IBM) would have, because they have a robust, entrenched
AFS/DFS installation with TBs of data, and money isn't really an issue
for them when the alternative is expensive reliability.  But of course
neither is administrative complexity; some of the IBMers appeared to
like debugging DFS kernel code simply because it demonstrated one's
technical superiority...

(end of pathetic synopsis... don't go basing your business on the above,
for your own good)

-- 

       "Every man is guilty of all the good he didn't do."

                                       --Voltaire
Received on 1999-05-10 18:20:41