You didn't say in your original post, but I would guess that your server has a
large number of files on it.

What you have documented here is a good reason why there should be some sort
of "checkpoint" facility available when full restores are being run.
Unfortunately, there isn't, and what you have described is the way it works -
it will read all tapes, even though the files are being skipped.  This
apparently is because the decision as to whether a file is to be restored or
not is made by the client, not the server.  I do think that some time is saved
though, since you don't actually transfer all of the data, etc.

The amount of time that is spent prior to the client restore being started is
spent getting everything in order - sorting the files into the correct order,
etc.  prior to actual data retrieval.  This is so that the data will be
retrieved in the most efficient manner - tapes are only mounted once, and read
through only once.

I would think that a checkpoint facility for an "interrupted" restore is well
worth the effort - the information for the checkpoint would only be needed
until the restore is complete, and could then be thrown away.  There have been
a lot of stories documented here about full restores that get stopped in the
middle.  What do you think, IBM?

Jerry Lawson
jlawson AT thehartford DOT com
________________________Forward Header________________________
Author: INTERNET.OWNERAD
Subject: Full Restore, ADSM Server Questions
02-24-97 10:39 AM

I have a question about the process that ADSM uses for restores.  We are
using ADSM Server for MVS ver 2.1.10.  The Netware client in question is
using the 2.1.6 client.  The started the restore, and after about 10
hours, the mainframe had a problem mounting a tape, and adsm had to be
bounced (this is another problem, not the one I am askng about - that had
to do with the silos).  So now, when ADSM was restarted, we did a restore
with replace=no, so that we would not restore files already restored, and
hopefull save some time.  It came to my attention, thought, that even
though we were skipping alot of files, it looks like ADSM server process
mounted all the tapes anyway.  Why is that being done?  It seems like
there should be enough information in the database to determine if a file
should be restored or not.  Also, I would like to know the process that
ADSM goes through when a restore is started.  It seems like he downloads
the full directory tree, but he must be doing something else also since it
takes almost an hour and a half before it starts "doing anything", from an
external point of view (the person watching it seeing any activity).
Thanks for any information that can be provided.

Andy Carlson        andyc AT andyc.carenet DOT org
BJC Health System
St. Louis, Missouri