Best way to handle Multi-Location DR setup

[brock]

I am looking for some input on the best way to handle setting up a new Veeam installation in my environment. The environment looks like the below:

Primary Site:
- 3 x ESXi hosts on an All-Flash Array
- Spindle Array for backups
- Redundant 16Gig Fiber SAN for ESXi hosts and both storage arrays

Secondary Site:
- 2 x ESXi hosts on an All-Flash Array
- Spindle Array for backups
- Redundant 16Gig Fiber SAN for ESXi hosts and both storage arrays

Their is roughly 80 miles between these two locations, but we are fortunate enough to have a dark fiber between the two, so their is 10Gig interconnect between the two locations, so bandwidth will not be an issue.

There are production VM's at both locations, and I will also be housing VM Replica's at opposing locations.(The bulk of the VM Replica's will be Primary Site VM's replicated to the Secondary Site)

I have Veeam B&R Enterprise Plus, as well as Availability Orchestrator and will be installing the Veeam software on Windows Server VM's. I am looking at probably running Linux Repo's at each location and taking advantage of the "Hardened Linux Repo" that will be available in v11 so that backups can be made immutable.

What is the best way to configure this? Backup all servers from Primary Site and send a backup copy of everything to the Secondary site? Backup the servers resident to each location with the Veeam B&R servers resident to those locations and then send backup copies to the opposing sites? Some other method?

Are the two Veeam servers aware of each other? Could I manage VM Replica's created on B&R Server 1 with B&R Server 2 in the event that B&R Server 1 went down?

Just looking for some good direction on the most resilient way to configure this.

Thanks!

[PetrM]

Hello and Welcome to our Community!

Veeam servers will not be aware of each other but according to our best practices it's recommended to manage replicas from Veeam B&R which is deployed on DR site. In your case, both sites are DR locations relative to each other, therefore it makes sense to deploy Veeam B&R instance on every site. Also, you can efficiently manage backups from the remote site if you have a proxy server on the primary one.

From my point of view it's better to point backups to a local repository and send backup copies to a remote location in order to respect 3-2-1 rule or to run remote replica from backup. The local backups will provide better RTO as faster restore is expected when data traffic circulates within the same site whereas replicas allow you to spin up workloads on the secondary site with minimal possible downtime.

In regards to VAO, I'd recommend to think about the scenario which you're going to automate either orchestrating failover or restore.

Thanks!

[brock]

So should I initiate a backup from the B&R Server at the same location where the servers I'm backing up reside? Or should I initiate backups from the B&R Server at the opposing location?

...and to confirm, I will send the backups to the storage at the same site as the servers that are being backed up, and then the "backup copies" to the storage at the opposing site?

Depending on the answer to the first paragraph, I'd assume that either A) I should build the VM Replica's with the B&R Server at the opposing site and utilize the Backup Copy Data, since that would reside in the same location where the replica's are being built? or B) I would have to most likely build the VM Replica's from the production data, as there would be no other way to build the Replica with the B&R Server at the opposing location if the backups are performed with the B&R Server that resides in the same location?

[veremin]

Or should I initiate backups from the B&R Server at the opposing location?

This.

I will send the backups to the storage at the same site as the servers that are being backed up, and then the "backup copies" to the storage at the opposing site?

This.

Thanks!

[brock]

Awesome! Thanks for all the help thus far!

[brock]

Reviving this thread, as I think this will add some value, along with introducing a few more questions and pro's/con's to this configuration. I don't see a lot of customers running a setup like this, so it was mostly a learn as I go scenario; and while I have been running the above configuration for close to 3 years now and it has been a very robust way to handle a multi-site backup configuration, there are some things that I think could be improved upon. It is really nice to be able to run each Veeam recovery site independently of the other and not have to worry about spinning up a VBR server, populating repo's, etc. I have replicas for most VM's at opposing sites as well, so it is extremely fast/trivial to get a server up and running.

Something else that I don't believe was mentioned in this thread, but is easily overlooked, is to be sure and make separate directories for each server. I've read that some people simply pointed both servers at the root of a partition, and that will cause a mess of confusion for the VBR servers, as each will see the others backups. I went so far as to create a ProdBackups, ProdReplication, DRBackups and DR Replication folder on the appropriate servers so that things were very easily identifiable. Structure things well and keep it organized so that you have less chance of confusion when troubleshooting, as running two VBR servers and sharing repository resources can get convoluted really quick.

One of the biggest issues I've run into is the inability to properly do Immutable backups on the Linux Repo's. The primary issue being that sending Immutable backups to the same repository from two separate VBR servers is not supported/recommended. To get around this for the time being, we are sending a third copy of our backups to a Veeam cloud provider.

We are planning for some future reconfiguration/upgrades to our Veeam environment and I would really like to have Immutable on all of our backups and copies. What options do I have outside of A) Stepping down to a single VBR server or B) building out two repositories at each location so that each VBR server has sole access to a repository? Would separate partitions on each repository circumvent this?

I feel like Veeam is missing a chance at more $$$ by not providing guidance and a supported way for customers to run two B&R servers.