Excerpts from transarc.external.info-afs: 6-May-93 And who should get to
"vos .. Randall [EMAIL PROTECTED] (1578)
> I assume these thoughts are being considered by those with one hand on the
> DFS keyboard also.
DFS has two answers for this.
The first, and possibly the most direct, is the whole concept of
scheduled (nee ``lazy'') replication, that automatically keeps the R/Os
updated to within some time interval of the state of the R/W.
Additionally, cache managers collaborate, by keeping the fileset
(volume) image presented to applications to within some (longer) time
interval of the state of the R/W, and also by not moving backward in R/W
history.
The second concept is that you can incorporate privately-owned file
servers into a cell, where those file servers are defined as being owned
by some group. If you're a member of the owning groups for all the file
servers that house all instances (R/W and R/O) for a fileset (volume),
then you can cause even a release-replicated fileset/volume to be
released.
Craig
