View previous topic :: View next topic
|
Author |
Message |
eugene90621
New User
Joined: 25 Apr 2006 Posts: 4
|
|
|
|
We are moving our production processes to a new data center. Currently all (production and test) our HSM migrated data is on the same ML2 tapes. Need a solution. I'm aware of the ABARS process but this may not be the way we wish to go. |
|
Back to top |
|
|
enrico-sorichetti
Superior Member
Joined: 14 Mar 2007 Posts: 10873 Location: italy
|
|
|
|
Quote: |
I'm aware of the ABARS process but this may not be the way we wish to go. |
ABARS as the name says is for backup,
I do not see how it would help to manage migration
but...
You are asking for a solution, for what ??
You are moving the production, ok!
what about the test.
to express an opinion we should need more info on your setup
if You are moving only one of the two environment the quickies from the design point of view is to clone BIT by BIT the system
and duplicate the ML2 tapes
there should not be any problem as far as the backup,
only the development libraries should be there |
|
Back to top |
|
|
eugene90621
New User
Joined: 25 Apr 2006 Posts: 4
|
|
|
|
In our current data center we manage both production and test. Being that we are located in a region prone for disasters we are building another center that we will only use as our production environment. Our current HSM (1) manage both prod and test with one set (no duplex or tapecopy) ML2 tapes.
Now we need to separate that data and move/recover the production migrated data set to a production only HSM at the new center. The test data will remain at the current center. |
|
Back to top |
|
|
enrico-sorichetti
Superior Member
Joined: 14 Mar 2007 Posts: 10873 Location: italy
|
|
|
|
I do not know how many ml2 tapes You might have,
and sincerely I do not remember if there is a tool to selectively
copy objects from a ml2 tape to another ml2 tape,
then the copy must be reflected into the mcds for the new tape.
there is a good probability that the dataset could be sprad on all the ml2 tapes
and most of the ml2 tapes will need to be mounted for the copy,
there might be a lot of selective tape handling and clerical work
to find out where things are.
that' s why I suggested to clone everything,
and in the final environment, disable thru appropriate racf permission/un-permission
the recall of the datasets related to the other environment |
|
Back to top |
|
|
eugene90621
New User
Joined: 25 Apr 2006 Posts: 4
|
|
|
|
This appears to be my best shot. Any other suggestions out there ? |
|
Back to top |
|
|
enrico-sorichetti
Superior Member
Joined: 14 Mar 2007 Posts: 10873 Location: italy
|
|
|
|
Quote: |
This appears to be my best shot |
I am not sure if it is the best;
it' s just a donkey work of copying all the tapes,
simple, stupid-maybe, but it is certainly the less prone to errors
unless somebody fumbles with the tapes or drops them on the floor.
the only problem is that HSM might mess around with the ML2 tapes during the copies
the other alternative is to list the mcds for the datasets
migrated on the test system, recall all of them,
change the HSM parameters to leave the TEST data alone
IPL the test system in lpar mode
create a new set of parameters for the test system
and command migrate datasets in order to free some dasd space
if You have enough dasd space could be simpler as far as interferences with the production HSM
Still You would have to play around with the HSM setup
from any point of view, at a first glance, there might be the need of some kind of frozen zone
But You' ll have to elucubrate on it, You know Yur setup |
|
Back to top |
|
|
eugene90621
New User
Joined: 25 Apr 2006 Posts: 4
|
|
|
|
I should have mention this earlier. We have some 500 (3590) ML2 tapes with some 9 terabytes of test data on these tapes and about 20 terabytes of production. (we don't have 9 TB available for recalls)
Our new shop will be virgin using somewhat of a D/R scenario to build the prod environment . We intends to do data replication for our DASD storage which is no problem with data separation. We had intended to make the HSM separation prior to the move running two HSMs one for prod and one for test. We could then do somewhat of a peer-to-peer of the tapes to the new center. |
|
Back to top |
|
|
expat
Global Moderator
Joined: 14 Mar 2007 Posts: 8797 Location: Welsh Wales
|
|
|
|
NO, you cannot selectively copy datasets from Migrat or Backup volumes by way of RECYCLE or anything else. A reall bummer as I would have loved to have had that functionality.
It's not only the ML2 tapes that need to be considered, as the usual scenario is that backup versions of datasets are retained for anything from a few days to a couple of years after the ML2 version has been expired.
How often do you process the EXPIREBV at your site ? How many backup versions do you have that should have been expired already and do not need any further attention ?
Will the two sites be in communication with each other. You may wish to consider making the two sites one SMS & HSMPLEX and transfer the data once this has been implemented. XPPRC should do the trick.
You could use SMS and HSM to split the workload processing between dev and prod so that prod is not impacted by taking backups of development environments, but DO make sure that the prod environment does take backups of development source and load libraries etc. etc. This will save the cost of reanalysing and programming the lost development. Let's face it, dev data is spawned from prod data anyway so as long as the code libraries are safe, the data can be easily replaced.
The HSM plex scenario also allows the possibility of two way duplexing, with prod data duplexed in the dev HSM, giving you even more stability. |
|
Back to top |
|
|
|