>I don't know exactly the criteria ADSM uses to determine whether to backup a
>file or not...

In the Unix environment, at least:
ADSM examines the file's attributes such as size, modification date and time
(mtime), ownership (UID), group (GID), and compares it to those attributes
of the most recent backup version of that file.  (atime - access time - is
ignored.)

As an aside, note that things get "interesting" when users manage their
HSM files.  For example, a user decides to change the group affiliation of
all files in a directory containing old, large files, as in:
'chgrp -R SOMEGROUP .'.  The next incremental backup sees all these files
as having been changed, and goes to back them up again.  Because they are
old, their data is now on tape, which needs another tape drive and probably
hours to perform the backup - all incited by this little, innocent command.
Because ADSM is generic to numerous platforms, it operates by object name,
and so does not remember that this inode is the same one it has always been,
so as to simply update its datbase info rather than back up the whole object
(file) again.

   Richard Sims, Boston University OIT