DR Test This Weekend - Replication Only

[dannythake]

Morning/Afternoon all,

We are performing a disaster recovery test this weekend and would like some advise on some of the processes. I will outline the environment below:

Production:
192.168.100.xx Subnet
2 ESX 5.5 Hosts connected to SAN
10 VM's (For this particular test)
Virtual vCenter Appliance - Replicated to DR site by using direct host connection

DR Site:
192.168.200.xx Subnet
1 ESX 5.5 Host with Local storage - Added into Production vCenter
Domain controller already at site, with Sites and Services properly configured
Veeam console that controls replication

Replication Job setup:
All VM's are replicated to the DR site. All are configured to Re-IP, except the Vcenter

Objective:
We would like to failover all VM's, including vCenter Appliance, test that all Servers are doing as planned and then discard all changes with the exception of our Mail Server, which we would like to "Failback to Production". My questions are around the order of processes, and the behaviour of the vCenter when performing each stage.

Our plan so far (purely related to the Veeam side of things), is to power down the entire Production site. Drive to our DR site (only a few minutes away). Start the failover using the Veeam console, starting with the vCenter, and then followed by the Replica's. Test that all servers are performing correctly. Discard the changes of all VM's except one, then failback the mail server to the production site with all changes.

Our main worries are around vCenter. Can someone please advise on whether this is the correct process? Could anyone advise on the below questions:

* When we power up the vCenter replica at DR site, do we just need to manually re-ip it, then re-connect it to the DR host, then begin the failover on VM Replica's?

* How will Veeam handle this change in vCenter config?

* When failing back, what would be the correct thing to do with the vCenter? Power down at DR, Power up at Production, then begin failback?

If anyone can spot any flaws in our plan, could you please advise? Any help would be greatly appreciated

Thanks

Danny

[Vitaliy S.]

Hi Danny,

I'm not Gostev.... but here you go:

* When we power up the vCenter replica at DR site, do we just need to manually re-ip it, then re-connect it to the DR host, then begin the failover on VM Replica's?

Yes, that's correct, however I'm not sure that you will need to reconnect your host, as it should be still in vCenter Server configuration, right?

Also please be aware that the best practice would probably be a configuration when Veeam backup server on the target site is connected to the standalone ESXi host directly. In this case to do the RE-IP on VM replicas the connection to vCenter Server will not be required.

* How will Veeam handle this change in vCenter config?

Veeam actually doesn't need to know what IP address is used on your vCenter Server. If you have added it to the console via FQDN/hostname and your DNS service can resolve that name, then Veeam backup server will not see any difference.

* When failing back, what would be the correct thing to do with the vCenter? Power down at DR, Power up at Production, then begin failback?

Failback operation is available only if you start the failover operation from the Veeam backup console. In your case I would suggest to shut down the DR vCenter Server and power up the one you have on the main site (I assume you will have a connection to that vCenter Server from the DR backup server, right?). After that use regular replication jobs from the DR site to transfer latest vCenter Server changes back to production. Use replica mapping feature for that purpose.

If anyone can spot any flaws in our plan, could you please advise? Any help would be greatly appreciated

If I were you would use standalone connection to ESXi host on the remote site, this would eliminate the step with manual vCenter Server config tweaks and will give you an ability to use Veeam native functionality such as failover/fallback for all VMs, including the vCenter Server.

Thanks!

[dannythake]

Vitaliy,

Thanks for your quick response. I can confirm our target host for each replication job is connected as standalone host, not through vcenter.

Are you suggesting that as our first job we should failover the vcenter through Veeam console, and then begin the Production VM's failover?

Thanks for your help, its nice to get a little confirmation!

Dann

[Vitaliy S.]

Yes, if you have this direct connection available, Veeam backup server will send commands to power on VMs directly to the host. Use failover operation to start your VMs, and when you're done with the test (source site is up and running) do the failback for VMs you want to transfer changes back to production and use undo failover for VMs for which you want to discard the changes.

[dannythake]

Perfect, thank you Vitaliy!

One last question if you don't mind. When failing back the single VM, do we power on the VM at the production site first, or does Veeam handle that? The hosts will obviously be up and contactable.

Thanks

Danny

[Vitaliy S.]

You don't need to do anything with production VMs during failback operation, everything will done automatically by Veeam.

Please keep us posted on your DR test results. Thanks!

[dannythake]

Hi Vitaliy,

As requested, here are the results.......

SUCCESS!!

We managed to bring up our entire environment at the DR location, including a virtualised VoIP system using both ISDN and SIP. A total of 14 VM's (some with a few terabytes of data) were successfully started without issue, and only losing a few minutes of data.

Thanks for you help with my previous questions.

Danny

[Vitaliy S.]

Perfect, thanks for the update!