Dave,

We do weekly offsite rotation, mainly because there isn't enough time each day
to copy the previous night's data.  We use collocation by node and have about
120 nodes.  We do not use the storage pool copy  function of ADSM for 2
reasons, 1) it is recommended that the copy storage pool NOT be collocated - I
couldn't imagine how long restores would take in a disaster recovery situation
if we did not use collocation.  We are using MVS as the server and get only
about .5 GB per tape so there are ALOT of tapes.  2)  If we did collocate the
copy storage pool, there would be way too much tape reclamation processing to
do to keep the pool at a reasonable size.

We have implemented our own method of creating tapes for disaster recovery.  We
have  tape collocation turned on by node.  Day 1 of the week is Sunday night.
On that night we start a new tape for each node by marking all tapes in the
pool read-only on Sunday morning.  We run the incremental backups each night
for the rest of the week.  Our server is MVS and we use CA1 as our tape
management system.  Using CA1, we are able to find all the tapes that were
created by ADSM since the last 'cut off' (Sunday morning).  We then use MVS
DITTO to copy these tapes.  The internal tape label of the primary and disaster
recovery tapes are identical but the external label of the disaster recovery
tapes has a different 1st number.  The rotation of these tapes is handled by
CA1 by manually (via program, of course) setting vault codes, etc.  In our
disaster tests, when ADSM calls for a 5xxxxx tape, we mount the 6xxxxx copy of
that tape.  Because the internal label is 5xxxxx, it works just fine.
Tuesday-Friday we select tapes from ADSM with utilization (not % reclamable) of
15% or less and stack that data (using move data commands) on to that week's
active tape for that node in an attempt to consolidate the data on the fewest
possible tapes.  If you would like more detailed information about this
process, I would be happy to share it.