Dean said, in part..
> > numerous examples...
>
> Virus protection?  Unix hackers?  Software distribution?  These are
not viable
> examples.

I don't see any examples showing the current scheme to be
wonderful.  Here's another example:

Two days ago we had a disk fail.  We (IBM) replaced the disk and we
formatted it and brought up the system.  Then we started to think
about restoration.  The person in charge of backup was not available,
but would be in a short time.  A restore of that "file system" was
started after a while and it completed successfully.  ADSM came to
the rescue and went away victorious (we didn't have to use the
OTHER backup system).

But let's take a look of what might have been.  What if a scheduled
incremental backup had run (automatically!) before we started the
restore?  Mayhem.  The restore would have been very fast ... ADSM
knew the file system to be empty!  Sure, I can ask the restore to show
me "inactive" files, but which ones am I to chose?  I assure you that
all inactive files for that file system WOULD NOT FIT on that disk.

Certainly, it would be a help if ADSM would display the date that a file
was backed up and the date that it "expired" and maybe whether it
was updated or erased, ... so that a person could manually do a
"point-in-time-restore" ... but then ADSM *could* do it so much more
rapidly and without error.

- - - - -

Dean later suggests that the cost would be felt ... the price would rise,
so IBM shouldn't do it.  May I suggest that this is Version *1* of a
VERY important multi-platform IBM product.  If IBM stops
development here, the product will exist for a short while, but will be
replaced by better products and procedures.  It is simply a matter of
whether one considers "point-in-time-restore" more important than
other important deficiencies, and whether the enhancement can or will
be used (and therefore attract more customers).

I don't have confidence that any development shop can make
important changes in a vacuum, though they do more often than not,
and often for non-technical reasons ... but I digress ...

I'm very confident that "point-in-time-restore" capability can be made
an integral an important part of ADSM ... and at an expense in terms
of data base and storage that might be expected.  Done properly, it
would LOWER my per-user data base and data storage costs.

Again, thanks for all of the comments.  I appreciate that IBM listens!

cheers,

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