1. Just change the setting to colocate, and do nothing else. ADSM will then
start writing colocated tapes, but the old data will stay where it is - not
colocated. Over time, as the data is scratched, it will be rclaimed to new
colocated tapes.
2. Define a new pool, and use it only for the new clients you are adding.
There will be no impact to existing tapes.
Optionally, later on, you could do a MOVE DATA, and move the data to the new
pool, after you have reassigned an old node to the new pool. Note that
defining a new pool, and moving nodes to the new pool will not affect the
data that is in the old pool - it will stay there until something causes it
to move.
An option would be to do the MOVE DATA back to the DASD pool, and then let
the normal migration handle the move to the new pool. This would optimize
tape handling somewhat.
3. Define a new pool, and make it the active path for all clients (new and
old). For example - you used to have "DASD---->TAPEPOOL" as a migration.
Now you would have "DASD---->NEWTAPEPOOL" as your migration path. Also, you
set "TAPEPOOL---->NEWTAPEPOOL" as a migration path. Then set the high and
low migration limits down, and let the pools migrate. Note that if you have
a lot of clients on an individual tape, there will be lots of tape mounts
required to migrate the tape.
One note here - I am not sure how ADSM would handle the mounting of tapes in
the last 2 examples. For example, if a tape has data from 10 clients on it,
scattered (randomly?) around the tape, and you only have a few tape drives
(say 3 or 4), whether ADSM will make multiple passes of the source tape (one
per client???) and copy to the new tape, or whether ADSM will make only one
pass of the new tape, and then mount the appropriate output tape. Perhaps an
IBMer can clarify this one.....
Other comments..... Obviously, there are a couple of issues to think about
here. How many tape drives do you have available? If you only have one,
then the only way to separate the data is by doing the MOVE DATA commands as
suggested with scenario 2. Do you have multiple automated drives, so that
you can let the robot earn it's keep? Then scenario 3 might make sense - get
it done quick. Do you have manual drives and need manual intervention for
each mount? This is a different issue - you at least need to supply lots of
bananas to the operators.
Perhaps others can suggest other scenarios..... but I think only you can
decide which fits your environment best. All of the above should work. If
someone thinks they won't - then flame away - I have on my asbestos pants today.
:-)
Jerry Lawson
jlawson AT thehartford DOT com
______________________________ Forward Header __________________________________
Subject: Collocation implementation
Author: INTERNET.OWNERAD at SNADGATE
Date: 6/18/97 12:41 PM
What is the best way to migrate from a non-collocation environment to a
collocation environment?
We are thinking about defining a new storage pool with collocation and
bringing in one server at a time.
We are concerned about the impact on the number of tape mounts.
Any ideas?
|
