Coda File System

Re: coda in 'production' systems?

From: Matt Peterson <>
Date: Tue, 02 Jan 2001 11:35:54 -0700 wrote:
> First, may I say that coda looks really good, and thanks to all those
> who have written it (and are still working on it).
> I run a cluster of linux boxes for a research group in a physics
> department and I'm trying to work out whether we should be using
> coda rather than NFS. It certainly looks good but I'm unable to
> find anybody else who is using it for 'work' systems.
> My users are computational physicists who essentially spend their
> whole day logged in and if the system is down they can't do an awful
> lot. (On the other hand, we're not trying to run a bank or a
> mission-critical database.) NFS hasn't given us many problems that
> can't be fixed by a reboot.
> So, my question is, are there many people out there who are using coda
> for their work systems?
> Thanks
> John


I have done some research on Coda performance in an attempt to come up
with appropriate areas of usage.  I would be happy to share my results
if you are interested.  Basically, we have decided that Coda is ideal
for use when the filesystems is close to read-only.  Coda is usually
extremely fast in read-only application, but performance suffers
markedly on writes.  Home directories (depending on usage), server
hosted applications, and such are good places for Coda.

Another thing you should realize is that Coda does not use kernel level
threads... at least I don't think it does (yet).  This means that Coda
will not be able to balance CPU load across multiple processors.  This
is one reason (at least for us) that Coda is not ready for full
production environment.

As for stability, I have not hard evidence that Coda is not stable
enough for production environments -- though I still feel really good
that our Coda filesystem is frequently backed up.

Matthew Peterson
Sr. Software Engineer
Caldera Systems, Inc
Received on 2001-01-02 13:36:01