I have seen two somewhat different approaches to this at the two places I have
worked with ADSM.

A previous employer used ADSM for backing up a number of AIX systems. They
initially used mksysb tapes to recreate the operating system and the ADSM
client software. They then used ADSM to update files in the root volume group
that had changed since the mksysb tape was created and to recreate files
outside the root volume group. ADSM was unable to update some of the files in
the root volume group because the files were in use, but these failures did
not seem to cause any difficulties. That employer later switched from mksysb
to SysBack/6000. All of this was several years ago. I don't know whether they
have continued to use this disaster recovery scheme with more recent AIX
releases.

My current employer takes a different approach. They reinstall the operating
system and the ADSM client software from distribution media and then use ADSM
to restore application files, other software product files, and a few
carefully selected system files (notably /etc/passwd and /etc/group). They are
very careful to install the operating system at exactly the same release and
maintenance level as the operating system on the original system.

We have found the DRM option very helpful in expediting the recreation of the
ADSM server.

Some software products that run under Unix require license codes that depend
on the serial number of the system they run on. Such products tend to be a
major nuisance during disaster recovery tests. We have generally been able to
obtain a list of system serial numbers from our hot site vendor before each
test and use these to obtain matching license codes from software vendors. We
are still working on procedures for obtaining license codes on an emergency
basis in the event of a real disaster.

We generate offsite tapes by writing them in our primary machine room and then
sending them to a vault (as opposed to having an automated tape library at
another location). This kind of hardware leads to some difficult tradeoffs.
Tape management schemes that require small amounts of tape activity to keep
the offsite tape pools up to date tend to require outlandish numbers of tape
mounts during disaster recovery tests. Schemes that require small numbers of
mounts during disaster recovery tests tend to require outlandish numbers of
tape mounts for keeping the offsite tape pools up to date. It has been very
difficult to come up with a workable compromise between these two extremes.