Protection Domains and Recovery in a disaster

  • 4 May 2016
  • 5 replies

Badge +6
Thoughts I'd post this rather than bugging our SE over and over ;)

Trying to get a disaster recovery plan in place to protect against a major disaster such as server room fire...

We have a couple of clusters geographically separated which I have set up as remote-sites. Site A and B lets call them.

I've set up a PD w/ consistency group with the VMs to protect and periodically snapshot from site A to B. Now in the case of a fire/destroyed cluster at site A... obviously step one is rebuilding a cluster at that site with the appropriate configs, but then...

With a brand new cluster in place, what is the standard process to retreive the snapshots from site B back to the new site A ?

So far, I've activated the backup on site B (as if it was an unplanned outage and left it off). and recreated the remote-site link on the newly built out site A. I am thinking I now create a new PD to shuffle the VMs/consistency group back to Site A and then activate there and I am done ?

I know this works as I've done it now. But is this the 'preferred method'?

I'd appreciate any thoughts from the gurus out there ;)

thanks, Brent

Best answer by Jon 4 May 2016, 18:42

View original

This topic has been closed for comments

5 replies

Userlevel 6
Badge +29
Basically you remake the PD with the same name, and that will allow you to pull the PD/snaps back
Userlevel 6
Badge +29
(Standard Disclaimer) if this were to happen in real life, always make a support ticket before doing anything, and we'll help point you in the right direction.
Badge +6
ha! perfect thanks for the pointer, I completely missed that
Userlevel 1
Badge +2
Could you pre-build the failback PD? No scheduled replications etc. but then after you migrate to the failover location it might allow expedite of the failback?
Userlevel 6
Badge +29
- hrmm, I dont think I understand the prompt there.

If you had 2 clusters, A and B, and you migrated a PD from A to B, and cluster A still existed, then by definition the PD would "be there" with some seed data, allowing you to re-replicate and fail back much easier.

In the case where cluster A "went away", then you can't really pre-build anything.

That said, If I missed the mark on understanding this, feel free to restate and I'll take a second swing.