[OpenAFS] Re: read-only to read-write fast?

[Benjamin P Myers]

On Monday 18 August 2003 03:45 am, Hein Roehrig wrote:
> tino.schwarze@informatik.tu-chemnitz.de (Tino Schwarze) writes:
> > On Sat, Aug 16, 2003 at 05:51:12PM +0200, Hein Roehrig wrote:
> >> Ideally, each workstation would become a AFS fileserver for a few us=
er
> >> volumes.
> >
> > Why do you want to do that? It makes your cell a lot more vulnerable
> > because the server keys need to be on each client. What are you tryin=
g
> > to accomplish by making clients a server?
>
> Save $$$ for a fileserver... In our current setup there is no central
> big fileserver and users tend to have their home on "their"
> workstation so that the NFS automount most of the time just does a
> bind mount and gives local fs performance. The obvious downside of
> this approach is that (total) disk failures tend to destroy a day's
> worth of work...

I had thought of the possibility of using lustre (www.lustre.org) for my=20
/vicexx filesystems. I have no idea about the usability of lustre itself,=
 let=20
alone the possibility of afs on top of it.

It looked promising because of a thread I saw in the archives that sugges=
ted=20
that it is possible to use NAS boxes as /vicexx over nfs.

Obviously, I haven't gotten this far as I'm just getting a handle on=20
installing openafs for the first time.  But I am, too, trying to think of=
=20
ways to harness the hundreds of gigs out there that we can't use on=20
workstations with more disk space than they need for the OS.

-Ben