Offsite vCenter server for DR?

[jabberwalker]

Hello,

I was recently at a conference where an administrator was going over their recovery plan and they mentioned that they made sure that their vCenter server was offsite because when attempting to restore an entire cluster, Veeam must be able to "talk" to vCenter. In my environment the vCenter server is on my primary cluster. Is this accurate? How would this change my DR planning? Should I move the vCenter to another cluster off site?

Thank you!

[Gostev]

Hello, in general vCenter is optional with Veeam, as you can always connect backup server directly to ESXi hosts and perform the restore this way, should vCenter become unavailable during a disaster. I would be more concerned about your entire vSphere environment, and how well will it function without vCenter. Thanks!

[bdufour]

i would have vcenter, veeam, and at least one domain controller installed at the dr site.

gostev - if you map veeam jobs directly to a host and not vcenter, and you vmotion the vm to a different host wouldnt that affect the jobs? would you have to remap them to the new host the vm has been vmotioned too? (general question, havent tested this, just curious how that would affect veeam jobs)

[DeadEyedJacks]

Hi,
One school of thought is to have a second vCenter at the DR site, so no recovery of a lost primary vCenter required.

Alternatively, you tie the vCenter VM to a particular host and target its backup job at that host, this then makes restoring the vCenter at DR site simpler.
Once you've restored the vCenter, then you can use it to restore all the other VMs which have backup jobs which targeted the vCenter.
(Think this is what Gostev mean't and avoids the vMotion issue you noted)

[foggy]

gostev - if you map veeam jobs directly to a host and not vcenter, and you vmotion the vm to a different host wouldnt that affect the jobs? would you have to remap them to the new host the vm has been vmotioned too? (general question, havent tested this, just curious how that would affect veeam jobs)

Yes, you would need to repoint the job to a new host in this case.