[OpenAFS] Questions about OpenAFS "reality"

[chas williams - CONTRACTOR]

In message <20060120214225.GI25719@riker.isk.kth.se>,Jan Johansson writes:
>All of the above have one common problem. When the disk run out
>you have big problem. "Ladies and gentlemen can we please have
>your attention. Please stop working, close your files and hold
>you hands up while we move your data to a new disk. Thank you." I
>am sure that this can be solved with NAS, SAN or some other hype
>technique of the week. But with AFS this is not a problem, you
>move user data under the feet of the user and they will not even
>know it. That is pure power.

to be fair, nfs never was intended to really solve this problem.  nfs just
exports your existing storage.  solutions to this "my disk is too small"
problem have been around a while.  ibm's jfs and lv are a good example
(and dates back to around 1991).  you can simply add new physical
storage to an existing logical volume and grow the filesystem to use
this new space.  since then quite a few others have added these features.
veritas has been around a little longer, but it has always been an extra
you had to buy.  irix, solaris and linux all now have similar features
to ibm's lvm/jfs.  there are some limits with this though--typically a
16T barrier somewhere.