Deployment Advice

[rlewin]

I'm deploying Veeam in our environment and am looking for some advice. My environment consists of two data centers that are less than two miles apart. These data centers are well connected with fiber SAN and 10G ethernet. I have a single VSphere cluster, consisting of two hosts at each site and SAN storage at each site. VMs that are stored at site A have an affinity to run on the hosts at site A and a similar setup for VMs at site B. I also have separate SAN storage for Veeam repositories at both site A and site B. This storage is not presented to VMware (not on the HCL). There are currently about 100 VMs and 25TB of data. I'm trying to setup Veeam so that if one of the data centers were to go down, I could restore essential machines to the other data center. Here is my deployment thoughts and I was hoping to get some feedback / ideas.

The Veeam server itself would be a VM. The database would sit on a separate SQL server (VM). Each site would have a physical proxy server running Server 2016 with a ReFS repository on the repository SAN storage at that site. VMs that are housed at site A will be backed up to a repository at site B and vice versa for site B.

First, does this scenario make sense? Does it fit my design criteria? Is there a better design? What happens if the VM that is the Veeam server is running at the data center that goes down?

TIA,

Rick

[foggy]

Hi Rick. Basically, your design answers your requirements. In case the site with the Veeam B&R instance dies, you will be able to fire up another Veeam B&R, import Veeam B&R configuration (you're going to back it up, right?) and perform necessary restores. However, for faster recovery, you could set up two instances (one per site) making cross backups (or replication, for even faster automatic failover) from another site (pull scenario). Also, keeping some short retention of local backups for operational restores is always recommended.

[rlewin]

I hadn't considered running two servers but it totally makes sense. I came up with a slight variation of this idea. What if I presented the SAN that is going to be used for the repositories to VMware and virtualize the entire thing: virtual Veeam servers and virtual proxies? My ESXi hosts use so little of the CPU that it shouldn't be and issue.

[foggy]

Virtual Veeam B&R servers and virtual proxies are totally possible, but storing backups inside a VM is not recommended.

[rlewin]

Thanks again for the advice.

[DaveWatkins]

You could also replicate the B&R server VM from site to site so if the worst does happen, you can just power up the replica