TSM disaster recovery with protectier vtl

Houssam

ADSM.ORG Member
Joined
Aug 19, 2016
Messages
12
Reaction score
1
Points
0
PREDATAR Control23

hello,
i have 2 sites with vtl replication (ts7620), in the site n°1 i have a tsm server 7.x.

i want to know how vtl replication and tsm work together in order to have a disaster recovery plan, i think that we have to choose between tsm replication and vtl replication, and not both.
 
PREDATAR Control23

i think that we have to choose between tsm replication and vtl replication, and not both.
Correct.

If you do VTL replication, you would need HADR to replicate the DB changes to the DR server. Or have an offline DR server that you would restore the DB too.

Or use TSM replication and deduplication, but your storage would need to be disk, not VTL.
 
PREDATAR Control23

dear marclant,

i do have an offline server to restore a db make a copy of volhist and devconf.
my question is :
if i have a vtl on primary site named ts3500-prim with 10 cartridges labaled primsite001 ... primsite010, and i have a vtl ont secondary site named ts3500-sec with 10 cartriges.
do i have to configure cartriges with the same label on the secondary site ? if so, do vtl replicate cartridge primsite001 of primary site with primsite001 cartridge of secondary site ?
how tsm knows about replicated backups
 
PREDATAR Control23

do i have to configure cartriges with the same label on the secondary site ?
They have to be the same name, TSM will look for known tapes labels (real or virtual).
if so, do vtl replicate cartridge primsite001 of primary site with primsite001 cartridge of secondary site ?
I would assume so.
how tsm knows about replicated backups
It doesn't. It knows about all the volumes it has used, if those volumes are replicated, the DR server can access those volumes in the DR VTL. TSM has no clue if these are the original volumes or replicated volumes.
 
PREDATAR Control23

If possible,
I would define second VTL on protecTIER @site2 and present to site 1 and do the db backups there ....
Also if servers are lunux/unix ... NFS from site 2 , export NFS and mount to TSM in site1 and do the db backups, volhist and devconfig files ...
 
PREDATAR Control23

Hi,
First, since this thread is quite stale I wonder how your solution looks now.
I'm working on similiar architecture right now, it's still not ready, but here is what I've found regarding PT and TSM.
HADR seems to work fine... but:
- additionally you need to replicate a lot of encryption related files located in "instance" directory. Otherwise "resurected" instance will not be able to autheticate virtually anything.
- turn on LOGINDEXBUILD on db2's. Even with that enabled, it can take long time to faliover.
- PT replicates cartridges. No need to create any on DR side.
- Replicas can appear in your DR site in two ways: on the shelf, shared between all defined VTLs, or in io slots of selected VTL. It depends on "visibility" setting of replication policy.
- Visibility is important - With "on" it triggers replication once the cartridge is ejected on the source library. And it appears in io slot of the target library. It makes you life easier with testing. Or is ecellent if you want to have offsite copy pool. With "visibility" off, PT replicates your cartridges more or less instantly, but they are not available in DR site untill you put it in "DR mode". Then you can find them on the shelf.
And here goes my doubts:
- If primary site goes down, both TSM and PT, on DR site I have my tapes, but the library is completly different. I'm working on clean solution of redefinig the library under the storage pool, and checking in the tapes back.
- I want to make it as automatic as possible, because I have 64 drives available lanfree to 10 hosts... That makes 705 paths to update, including TSM :-\
I'd appreciate any sugestions how to design and automate such solution.
 
Top