We have implemented a plan for Disaster Recovery of our most critical
client servers. Our ADSM server(2.1.3) is MVS. Our objective is to recover
critical fileserver(s) in a minimum amount of time which means as few tape
mounts as possible.
We created a "DR" node (IE, FS020DR) for each critical client server. It
actually is set up on a different schedule. Below is the steps we have used
to set this up.
1. Defined new Domain, Policy Set and Management Class.
2. Defined a copy group. Initial backup will back up all files, so we
basically started with a full backup without copying old data over. This
allowed us to retain all our daily backups, our weekly DR backups will only
have the most current version of a file. Of course, you can choose to
maintain as many backups as you want.
3. Set up a separate primary storage pool with a copypool. The tape primary
pool is set up with collation. The copypool is not set up with collocation
and will be sent offsite.
4. Created new DR nodes for the selected fileservers. ( Which means each
client server has 2 nodes). This had to be done in order to keep the same
management policies on daily backups. An ADSM node can only be assigned to
one Policy Domain.. We wanted to be able to restore active versions of files
without a large number of tape mounts.
5. Defined a new schedule for the disaster recovery pool and associated new
DR node.
6. On the client server side a second scheduler was started with the
-NODE=newDRnode in the parameter setting.
7. Allow access so the original node can restore from the new DRnode backups.
7. Allow access so the original node can restore from the new DRnode backups.
We have backups from both nodes (IE FS020 and FS020DR). We can actually do
a full restore with our DR node with minimal tape mounts, then apply
incrementals from the FS020 node.
This doesn't exactly fit your criteria, but you may be able to use some of
this to implement the requirements for you environment. Hope it helps.
At 08:23 AM 8/1/96 -0400, you wrote:
>---------------------------- Forwarded with Changes ---------------------------
>From: INTERNET.OWNERAD at SNADGATE
>Priority: Low
>Date: 7/31/96 6:22PM
>To: Jerry Lawson at TISDMAIL
>*To: *Multiple recipients of list ADSM-L at SNADGATE
>Subject: Implementing collocation?
>-------------------------------------------------------------------------------
>
>Jeffrey -
>
>I don't think I have a complete answer for you, but I will start the responses
>rolling with my own experience here - perhaps someone else can fill in the
>gaps.
>
>We did something very similar to what you did, although it was in the other
>direction - we had everyone collocated, and then chose to have our PC backups
>not collocated, due to the number of nodes, and the number of potential
>partially filled tapes associated with that. So we split into two pools, and
>kept the servers collocated, etc.. However, since this was not (at the time)
>covered by a disaster recovery pool, we did not try to move data from pool to
>pool - we just left it there. :-)
>
>> 1. Is there some way to move the _existing_ files in ADSM
>> storage pools for these servers into the new collocated
>> storage pool?
>
>Since you are not collocated to begin with, I am not aware of any way to do
>this that you haven't already mentioned. MOVE DATA works on a volume basis,
>and thus doesn't help in your configuration. The only thing I could think of
>was an Export/Import operation.
>
>> 2. Is the separate policy domain the right way to do all this,
>> or should I be reorganining this at a different level of
>> the storage hierarchy?
>
>IMO - you have the right approach. Actually, I can't think of a different
>way, but it is early. :-)
>
>Jerry Lawson
>jlawson AT itthartford DOT com
>
>
>
>
>
>________________________Forward Header________________________
>Author: INTERNET.OWNERAD
>Subject: Implementing collocation?
>07-31-96 06:22 PM
>
>Because we're implementing offsite backups for 4 of our 80
>ADSM nodes, using EXPORT NODE on a VM Version 1 server
>(*please* IBM send ADSM Version 2 for VM!), we get a _lot_
>of tape mounts when we do the export, although at the
>latest server PTF level, the run-time is otherwise OK.
>
>I'd like to implement collocation for just our four RS/6000
>servers, leaving all our desktop systems with collocation
>off to minimize tape mounts during our daily backups. I
>have created a new policy domain for our servers, including
>a separate storage pool with COLLOCATE=YES. When I switch
>the servers to the new policy doman, their backups will
> start going into the new collocated storage pool. Questions:
>
> 1. Is there some way to move the _existing_ files in ADSM
> storage pools for these servers into the new collocated
> storage pool?
>
> 2. Is the separate policy domain the right way to do all this,
> or should I be reorganining this at a different level of
> the storage hierarchy?
>
> Jeffrey A. Brooks (304) 293-5192
> IBM Systems Group jeff AT wvnvm.wvnet DOT edu
> West Virginia Network for Educational Telecomputing
> Morgantown, West Virginia 26505
>
>
|
