[OpenAFS] ...Those That Help Themselves

[Andrei Maslennikov]

On 30 May 2001, William Setzer wrote:

> As such, my boss wants to me to feel out the other Universities and
> businesses.  Have you thought about what's going to happen to AFS in
> three years or if it'll even exist then?  What about a possible move to
> OpenAFS?  What are you going to do about it all? 
> 

Hi William,

I believe the move to OAFS is exactly what may happen in a close future. 
We have deployed several OAFS cells in university departments, on Linux
(RH7.0), including database, backup etc, and everything works just fine,
for months. Cheap and beautiful.

Now, consider this: one of our main cells is based on Solaris 8 machines,
with fibre channel raid arrays and tapes. We use Solaris (and Solaris 8)
mainly because of better hw support for FC. But see what happened to
official AFS after OAFS source code snapshot was released - at the moment
we are obliged to run this incredible mix of binaries on sun4x_58
machines:

   - kaserver sun4x_57 build 2.3 (otherwise IBM SP k4 administration 
                                  does not work with AFS)

   - buserver/butc sun4x_58 build 2.6 (otherwise backup db is getting
                                       corrupt)

   - all the rest sun4x_58 build 2.13 (backup command now correctly works
                                       for "add/delvolentry", but for
                                       "volinfo" 2.3 build has to be used)

(Maybe 2.18 available as of yesterday will address some of these issues).

I.e., while the official builds trend to become more buggy (or absent, as
in the case of Tru64 5.1), OAFS is gaining momentum, especially on Linux. 
So Linux/OAFS is becoming a winning combination. We have just finished
first series of tests of fibre channel SAN connectivity on Linux, and
results are quite promising. And we now have plans to stress and benchmark
our first OAFS server on Linux with fibre channel periperals. If it works,
we may jump the Linux/OAFS train also on the principal cells. 

So IMHO OAFS is very serious, especially on Linux (because of the large
user community). To convince yourself, maybe start with an OAFS-only
stand-alone cell on Linux servers.

Andrei.

____________________________________________________________________________
Andrei Maslennikov                                  phone : +39   06 4463354
CASPUR Inter-Univ. Computing Consortium and INFN    cell  : +39  335 6214776
c/o Universita' "La Sapienza", 00185 Rome, Italy    fax   : +39   06 4957083
___________________________________________________ mail  : andrei@caspur.it