[OpenAFS] OpenAFS vs NFSv4?

[Mitch Collinsworth]

On Mon, 28 Apr 2003, Rodney M Dyer wrote:

> At 12:00 PM 4/28/2003 -0500, you wrote:
>
> >Well, though not as clean as AFS, the clients don't have to connect to
> >a single network end-point. Any NAS head in the cluster can service the
> >the client, even if the information is not on that NAS head.  And, if the
> >back-end storage is on a SAN, then any NAS head can cover for a failing
> >NAS head. Basically they've put the "AFS client code" on the heads instead
> >of the individual clients.
>
> That's just great <sarcasm>.  It's always so good to know that I'm doubling
> my network bandwidth by using a gateway product...  Or, even better,
> needing to add yet another network infrastructure to support the back-end
> SAN network.  No wonder IT costs are going through the roof.  No one wants
> to invest in doing it right the first time, and developing "infrastructure"
> around it.  Things should get easier as we go along, not -more- complex.
>
> Rodney

Hmmm...  It's been some months since I looked at it but I'm not fully
convinced their system would not result in a lower total cost to operate
than an AFS cell.  There might be scaling issues for very very large
cells, I don't fully recall the details.  But I think most of us would
agree that while AFS make a lot of tasks easier for the sysadmin than NFS,
it also makes some tasks more complicated.  If this product can leverage
the best of both and actually be easy to administer, then it might
actually come out ahead in the end despite the purchase price being higher
than a bunch of Linux servers and low-end raid boxes.

I'm not saying go buy one tomorrow, but I do think folks that like the
features AFS adds to their environment should at least give it a look.
If nothing else, give it a look and write to Spinnaker and tell them
what they'd have to add to it before you'd buy it.  They are very
interested in hearing what the AFS community thinks of their product.

-Mitch