Dean said..
> Point-in-time-restore = ARCHIVE

First let me thank everyone who has responded on the list and
privately.  I am encouraged that (some) others share this concern and
think enough of ADSM to participate in these discussions with the
hopes of learning more about the current product and providing input
to the developers in the form of our concerns.  (I'd rather provide the
input in the form of code, but ... :-) ...)

Archive is OK for some things.  For example, we had a request to
save the contents of a person's disk and restore it in about a year.
That is an archive function.  It is not what I meant by
"point-in-time-restore" in my original article, though.  This is a single,
special-purpose, one-time event.  An archive.

"Point-in-time-restore", as I meant it and as several people have well
described it, is a recurring periodic BACKUP function that ADSM, IMO,
needs.  (It needs a few other things, but this is by far the most
important to me and the University of Maine System).

Some of the posters have suggested that "point-in-time-restore"
doesn't "fit" with the ADSM structure, yet I think that is probably not
true and, at least from my perusals of ADSM-L logs, has no basis in
fact.  It has been suggested in private responses to me that this topic
has been discussed significantly within IBM, and though no work has
been done or is being done, there may be interest in ADSM
development.

I can visualize point-in-time restore from a user's perspective ... we
just need to see directory structures as recorded at various backups.
These are probably already in the server, we just need the client to
expose them and for the server to expire files and directories in a
quite different way.

Please keep those cards and letter coming!

cheers,

Wayne T. Smith
Systems Group -- CAPS        internet: wts AT maine.maine DOT edu
University of Maine System   BITNET/CREN: WTS@MAINE