Deployment Advice

[sixarm]

I'm in the process of standing up a new DR site using Veeam as our backup and replication solution. We have purchased two physical servers with attached storage to act as Veeam repositories, one at primary and one at DR. At the primary site, the physical server will have access to our 10G network to pull storage snapshots from our NetApp for certain VMs.

Having not deployed Veeam in production previously my question is what Veeam roles are best to deploy to the two physical servers? I know there are many ways to skin a cat, as it were, so I'm not looking for a definitive answer, I'm more interested in peoples thoughts and to spark discussion.

The Veeam documentation suggests installing B&R at the DR end so that it's always available in the event of a loss of production, which makes sense. If I went down that route, what roles would be installed at the production end?

I look forward to hearing your thoughts and thanks in advance.

[Deon]

Hello sixarm,

If I understood your setup correctly, you would want something like this. I am not great at diagram drawing, but hopefully it is clear enough.
You may also want to have a few virtual proxies on the hypervisor VMs for Hot-add operations (i.e. target write for replication if you plan to do on-site Veeam replication; replica disks cannot be written in Direct-SAN mode).



Of course that's just a generic picture, so if you have a specific setup in mind, please share.

And it goes without saying, you would want a connection with necessary ports open between VMware and 2nd host, to be able to restore from DR Repository to production in case your 1st host is temporarily unavailable.

[sixarm]

Hi Deon,

Thank you very much for the reply and for the diagram, that's amazing!

I've been doing a bit of reading around this and one thing that came up was that using a single B&R server in DR would mean the loss of Production backup jobs if there was a network issue between Production and DR. The recommendation here would be to have two B&R servers, one at Production and one at DR. The Production B&R server handles backup jobs, and the DR B&R server handles replication. That way, if you lose network between the two sites at least your Production backup jobs continue and replication would fail all the same.

I like the sound of this, albeit with a slight management overhead with the two servers.

Are there any caveats around this?

A few queries I have:
1) - If you lost Production completely would the DR B&R server have visibility of all data within it's repository to allow you to restore to DR?
2) - Would Backup Copy jobs need to be configured on the Production B&R server or would the DR B&R server have visibility to allow them to be configured there? For clarity, to copy backups from Production repo to BR repo.
3) - Is this where Enterprise Manager comes in to play?

[foggy]

1. DR backup server will be able to perform failover to replica VMs. Also, you will be able to import backups created by production backup server and perform restores, if required.
2. If backup copy is implemented by production instance,you will be able to import backup into DR instance. To implement backup copy on the DR instance, you would need to use a dummy backup job, specify it as source for the backup copy job and perform source repository rescan prior to each copy job cycle.
3. Not necessarily.

[sixarm]

Thanks for the rapid reply!

I'm not sure what to do now, I feel like I'm just massively over-complicating things! Any further pointers?

In reality, we very very rarely lose network connectivity between our Production and DR sites, and I really like the simplicity of a single B&R server at DR with a proxy and repository at Production, as Deon drew out. I think I'm swaying towards that at the moment.