DR Testing Procedure

[P Arumugavel]

Hi Gurus,

We have a plan for doing DR test in our environment.

Do anyone have the plan template, can you please share with me?

Also I just want to know how to move the data from the PROD to DR, I know the disks are presented in PROD are replicated with DR storage. So is that enough to do only vgexport and vgimport? If so, how is it happening since the disks are hosted from different storage... :( I have never done the DR test)...

So more information about DR test  would highly be Kudoed...

Thanks....

\ Vel

[Bill Hassell]

Dependiong on whether you have a simple one disk server, or several dozen large scale servers with hundreds of LUNs, the plan might be one page to hundreds of pages of procedures. Syncing data between PROD and DR is almost always dependent on how much the data changes and the maximum amount of money your company is willing to spend on high speed links. Disks are many, many times faster than long distance links and depending on the acceptable outage time, you may find it cheaper to just make tape backups than try to synchronize the two machines.

Without a description of the two machines (model, disk space and array model, network speed between PROD and DR, etc), only primitive recommendations can be made.

[rveri-admin]

Hi Vel,

If you are going for a DR test ,  you have to have a solid documented procudure,  can be of numerous page long., depending on your enviroment.

You can do a virual DR test also, depending upon your network & server infrastructure.

If you are planning from Site A  ( Pri)   to  Site B (DR)  ,  recovery test, make sure few things are replicating every moment.

If you dont have a direct replication service like storage replication, you need to depend on Tape(s)/DAT/DDS etc to be able to transfer primary data to DR site and do the test.

If you have continous access XP, replication you can do it well and fast and with ease. Such cases your DR environment should have similar type of cluster environment as the Primary site has. With similar package configuration under MC/Service Guard environment.

1. Split Continous access replication.

2. make VG's cluster unaware , make cluster aware in DR site host.

3. veryfy the IP's of the packages, (You can have different set of IP's than PROD Ips.

4. bring up the packages.

5. Have DB/Apps team check.

6. Carry on Test.

7. Done.

8. Bring down applications, bring down packages.

9. Check control file log. /etc/cmcluster/pkg/pkg.cntl.log

10. Resync the storage replication. make sure all came to PAIR state.

11.Done.

If you describe more about the type of environement you have, with OS/Cluster/Storage/Backup/Repliation etc, it may be easire to explain more.

Cheers,