Hi Ron,

> Can someone explain how ADSM's data base works.

The data base doesn't hold file data, it holds various kinds of
information, dominated usually by information about each file version
backed up.  An estimate I've used successfully is 800 bytes of DB per
file (1600 if you use mirroring) with some substantial portion left
over for room to grow.

So, you can predict, more or less, the tape costs of a 2GB server (for
example, for 3490 tapes each tape holds a minimum of 200 MB),  but to
predict the DB costs, you need to know the number of file versions
backed up.

> This is only one of several servers we want to backup, not including
> all of our workstations. We can't dedicate that much DASD to ADSM.

ADSM resource cost is dominated, at least in my VM server case, by
data base cost.  It took a while for us to figure out that the first
copy of the client disk is very important to have, but multiple copies
are not worth the (ADSM DB) cost.  ADSM probably doesn't treat your 7
copies of each file like you'd want anyway.  Search for
"point-in-time-restore" in the ADSM-L archives for some discussion on
this area.

cheers,

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