[OpenAFS] proactively replicating volumes

[Jacob Gorm Hansen]

hello,

from the books I have mentioning AFS, it appears that only manual replication
of volumes is possible - ie you have to manually release the volume for your
changes to be forwarded to other servers. Is this correct? If so, do people 
use cron jobs for this kind of work? I guess a small lag in replication would
be ok for my uses, though pro-active replication would be much nicer.

It seems to me that making AFS just push all changes from the master RW volume
to a number of slaves would be simple to implement (so perhaps it already is).

I would like to create an environment with one large and expensive server,
being replicated by a number of cheaper servers, so that if the large one is
down, one or more of the others will be able to take over with little or no
administrator effort.