Hi all,
I use SRM 5.1U1 and EMC Recover Point as SRA. We implemented SRM/Recover Point successfully.
We tested our SRM implementation lots of time by using a sample of about 20 VMs and a non-production consistency group :
We used "Test" successfully
We used "Planned Migration" successfully
We used "Disaster Recovery" successfully
We did the same successfully by cloning more than twenty VMs used by an application to validate its running after a recovery / failback.
Now, it's time for us to check our recovery plan on our Production...
Running a "Test" won't be possible as the SRA's caches are not big enough; when the caches are full, the VMs will be hung.
Running a planned migration on our DR site is not possible just for a test, the production must continue during the test.
I know that lots of people will tell me to launch a planned migration and then a failback, they would be right, but, for now it is not possible yet.
So we decided to try this scenario :
- Isolate Disaster Recovery site from the Primary Site by "cutting" WAN connectivity at the DR Site
- Let Primary Site runing the Production
- Launch a Disaster Recovery from the (isolated) DR Site
- Check the result (during several hours or about one or two day) so as to check if VM contents work normally
Then we want to "clean" the DR Site without inverting the replication (so no "Reprotect") before reconnect DR Site to the WAN.
How can we do that?
I insist on this point, the production site mustn't be impacted in its work.
We know and assume that datastore consistency will be wrong after the test, then replication from Primary Site to DR Site have to be started from the beginning; and the protection wouldn't be unavailable until complete replication.
I have read a lot of articles without finding any that correspond to this scenario.
Any suggestion?
Thanks in advance.