Exchange DR scenario / Restore single DAG node

[dbr]

Dear all,

I'm sorry to dig out a topic that was already discussed many times (e.g. here) but I cannot find a clear statement. How do I recover a failed Exchange DAG node correctly and is it possible to recover a whole DAG by restoring all DAG nodes? I'm afraid of crashing my DAG when restoring a single failed DAG node with full vm restore like it might occur for Active Directory unless you use vss to let Veeam do the work for you (fully automated non-authoritative restore).

Assuming one of three DAG nodes failed and will not boot up anymore. Is it possible to restore a DAG node by use of Veeam full vm restore without risking Exchange configuration or Active Directory corruption and the node cannot mount any databasecopies at worst? Or am I forced to recover a failed node by installing a fresh server with same configuration (ip, name etc.) and install Exchange with "Setup /m:RecoverServer" (Whitepaper)? I want to safe time and minimize effort and surely restore the whole DAG vm will be much more faster. Or in other words: Does Veeam provide similar recovery features for Exchange recovery as they do for Active Directory?

Daniel

[foggy]

Hi Daniel, here you can find some tips for DAG restore. And a couple of other existing threads:

Exchange 2010 DAG backup ?
Restore Exchange 2010 DAG

[dbr]

Thanks for your reply Alex. Unfortunately this doesn't answer my question exactly. Most part of the links are about backing up DAG. Only thing I noticed is the section "Tips for restore planning with Veeam" in the first link and the statement that Veeam does process some restore steps. But as I understand they are related to databases only!? Or what is done by Veeam in detail?

However, main question is still if it's really save (with regard to Active Directory and DAG) to restore a failed DAG node or a whole DAG cluster with Veeam independent from restoring Active Directory? I'm a bit afraid of doing this due to the statements of orangep7 in the first link from my opening post. Ok, I can test this in a virtual lab but I would have a better feeling if someone can confirm this scenario in production environment for a case of desaster. Answer to this question (Do I have to recover a DAG node with Exchange setup built-in recoverserver-switch or can I just use Veeam full vm restore) will affect much the way we design our Exchange recovery process.

Does anyone ever had to restore a failed dag node with Veeam (regardless of Exchange version) with success?

[hwelvaar]

Good question, I'd like to know that as well.
Has anyone successfully restored (a node of) Exchange DAG with Veeam ?

[dbr]

Even though this is a quite old topic I would like to share my experience. Because I didn't find anyone that had ever restored a DAG-node with Veeam, I decided to follow the Microsoft best practice. Our DAG consists of 3 Node, two of which are on dedicated ESX-Hosts and one is running in an ESX-Cluster with shared Storage. First two nodes are not backed up, only the one in the cluster with a combination of Veeam ordinary backup and Storage Snashots in order to have a backup and truncate Exchange transaction logs. If any host crashes (including the one backed up with Veeam), I will setup a new vm from scratch and install Exchange with the recovery switch and redeploy the database copies from the other hosts. I had to test this szenario because we had a raid set crash on one of the dedicated hosts. Because we have three nodes and the cluster was still functional with two hosts I had enough time to recover the lost machine by above mentioned recovery scenario. In our case Veeam is only used for item level recovery and in case we have a logical failure in Exchange environment at all.

[ejenner]

I think you choose a Veeam backup as your resilience, probably set it with a fairly short schedule, a couple of times a day (or whatever you want). Just backup individual Exchange servers.

Or....

You go down the DAG route.

Trying to do both at the same time is like trying to provide the same resilience twice so it's maybe not necessary?

With a DAG configuration you would use Veeam as a way to perform disaster recovery rather than recovery of a single node.