DebbieW wrote, in part..
> We initially implemented ADSM w/DRM employing the incremental ideology.
> However, the cost of having a full duplicate set of tapes offsite for data
> that is retained forever (in some cases) has become a strain on the storage
> media budget.  The powers that be have requested (actually, it was a bit
> stronger than a request) that I change our backup plan so that only the most
> recent "full" backups are backed up to copy pool tapes and sent offsite for
> DR.  We plan to continue our regular incremental backups for normal
> recovery, with the existing retention parameters, so the "full" backups will
> only be retained for 14 days.

Comparing the costs of full-backups vs ADSM's backup-once
incremental philosophy is very difficult and maybe the differences are
not obvious.  From the information given, I know neither your ADSM
nor full-backup situation, so let's take a simple non-DB example:

Suppose the full-backup is done weekly and kept for 2 weeks less 1
day.  The minimum number of tapes is 2 times what is needed for a
full backup.  This is an incredibly short time for most shops, but is a
minimum.

    One of my mainframe systems does a full backup weekly and saves
    for 40 days, plus an every 4 months full backup kept for one to
    two years.  Oops, I forgot this doesn't include DR ... that's
    another 3 sets ... for a total (each permanent file) existing on about
    12 tapes!  Many ADSM policies will require *fewer* tapes.

Now let's look at ADSM's method.  Everything is backed up to a local
tape and a DR tape (for example).

    Obviously, the DR tape is optional. The ADSM local copy MAY BE
    your DR tape, if it is offsite of the spinning disk, and depending
    on your DR definitions and requirements!

In a static environment, the ADSM and full backup techniques yield
the same number of copies on tape(2)!  Over time, the analysis gets
very difficult, but in general, your ADSM policies and procedures will
limit the tape usage.  You will generally have just two ADSM copies of
each object, no matter how many versions your policies allow.  So
your increase in tape usage depends on the changing of files and
thus the increase in backup objects.  The static objects are on exactly
two tapes and cost no more than the full-backup method ...  less if
more than two sets of full-backup tapes is used.

So the significant ADSM analysis should be made on the *changing*
files.  Policies may lead to many copies of changing files within
ADSM, and these ADSM tapes are, on average, not close to full, for
several reasons.  Keeping relatively full ADSM tapes means more tape
handling, and data movement at the ADSM server in exchange for
faster restores and fewer tapes needed.

If you did weekly full-backups and daily incrementals pre-ADSM,
translated that into retain-extra-versions==14, and have a highly
changing disk store, then ADSM will cost significantly more in tapes
and DB.  I don't have any easy solutions in this case! Maybe others
do. :-)

Your analysis for archive and DB backups is probably significantly
different and may need to be looked at separately from file backups,
much as you have various ADSM policies.

On the other hand, if your ADSM tape usage is significantly more than
you expect, you must look for the unexpected.  Is expiration really
running to completion?  Are there file systems that haven't been
backed up in a very long time? Do you have lots of "full", but little
"utilized" tapes? Do some systems essential do a full-backup every
night and have relatively high/long retention policies?  Others can no
doubt think of may other possibilities. :-)

cheers, wayne

Wayne T. Smith                          ADSM AT Maine DOT edu
ADSM Technical Coordinator - UNET       University of Maine System