I'll wade in here and hopefully the folks that really know will correct any
misconceptions I've been propagating.

I believe that when a file gets backed up or archived a reference to that
file is placed in to the ADSM database.  This reference includes a bunch of
information fields including:

File Name
STG Pool Location 1
STG Pool Location 2
STG Pool Location n (Since the client data can exist in many pools
simultaneously)
Backup/Archive Date
Management Class

There's probably lots of other stuff, but I haven't divined that yet.  Help
by the experts is appreciated.

So, part of the info stored is the management class name.  Not the
parameters in the management class, but simply the name.  If the parameters
of the management class change, for instance the retention period, client
data retention periods within that class change as well.  Nothing more
needs to be done.

Expiration processing does the work.  I believe expiration works as
follows: the process compares the date the object was backed up/archived
with the information in the management class and a decision is made to keep
the file or delete it.  There is probably some variation on the brute force
method I just proposed that optimizes this process.

Bottom line: by changing the retention period in your archive copy group
and activating the policy set (don't forget that part or you'll be right
back here) will change the retention period of your archives.

Blast away.

Kelly