Veeam architecture for DR site?

[kevinnaughton]

I've just read a dozen thread, with several asking the same question, but I haven't found a clear answer. Here's the scenario.

Running VMware essentials plus and Veeam Essentials.

Production Site:
- 2 host cluster. vCenter and Veeam server are VMs in the cluster. Cluster sized so that all VMs could run on 1 host.
- Backup repository located here (physical server)
- Veeam using local DB

DR Site (different subnet):
- 1 host, replica target.

What to do about Veeam servers for easiest DR recovery?

Option 1:
- Replicate Veeam server to DR. Recovery means manually bringing up the Veeam server replica and re-setting the IP. Then using Veeam to bring up the vCenter replica. Then bring up the rest.

Option 2:
- Second Veeam server at DR site. This seems like a better option as I will be able to bring up replicas right away. However, I can't find any information about how 2 Veeam servers share DB info.

Option 3:
- Veeam server in DR only. This seems cleanest option for quick recovery. However, I read that guest file restores in procution would then have to go over the WAN.

So, I'm thinking that Option 2 is the best. But although I've seen it asked several times, I can't find an answer regarding communication between the Veeam servers when using the local DB.

[kevinnaughton]

I think I just came up with the solution. One Veeam server in production used for backup jobs, and another at the DR site for the replication jobs. Then I don't need to worry about what the other knows, and I'm all set to bring up the replicas.

One question though, will I need to carefully coordinate the timing of these jobs? Some of the VMs will be running near-CDP. Will I need to build in a window for backups where those jobs are paused?

[Vitaliy S.]

- Second Veeam server at DR site. This seems like a better option as I will be able to bring up replicas right away. However, I can't find any information about how 2 Veeam servers share DB info.

Veeam Backup servers cannot share the same database, each backup server should have its own database.

I think I just came up with the solution. One Veeam server in production used for backup jobs, and another at the DR site for the replication jobs. Then I don't need to worry about what the other knows, and I'm all set to bring up the replicas.

Yes, that would work. Also I would suggest installing Enterprise Manager to have visibility across two backup servers.

One question though, will I need to carefully coordinate the timing of these jobs? Some of the VMs will be running near-CDP. Will I need to build in a window for backups where those jobs are paused?

Yes, if you have two separate backup servers and all jobs are going to use application aware image processing, then I would recommend using separate job schedules.

[kevinnaughton]

Ok, thanks. I'll see if I can figure out a way to script is so that the continuous replication jobs are paused while the backups run and then get re-enbaled. Unless someone happens to have that solution handy?

[Vitaliy S.]

Actually there is no need to disable job completely (if you go with two backup servers), the conflict may only happen during VSS freeze/unfreeze operations. You can just leverage "backup window" settings to configure allowed and prohibited hours for replication job to make sure these jobs do not overlap. See our User Guide (page 149) for further instructions.

[Gostev]

if I can figure out a way to script is so that the continuous replication jobs are paused while the backups run and then get re-enabled.

If you go with the single Veeam server option, then as of 6.0 or later, this will happen automatically when two jobs try to process the same VM at the same time (specifically to prevent VSS conflict Vitaly is talking about). So, you do not need to do anything at all.

[kevinnaughton]

Actually there is no need to disable job completely (if you go with two backup servers), the conflict may only happen during VSS freeze/unfreeze operations. You can just leverage "backup window" settings to configure allowed and prohibited hours for replication job to make sure these jobs do not overlap. See our User Guide (page 149) for further instructions.

I need to continuously replicate. A pause for actual backup processing is acceptable, but not a large window.

[kevinnaughton]

If you go with the single Veeam server option, then as of 6.0 or later, this will happen automatically when two jobs try to process the same VM at the same time (specifically to prevent VSS conflict Vitaly is talking about). So, you do not need to do anything at all.

But if I do that, I read that guest file restores on the production side will have to go over the WAN and back. Is that correct?

[foggy]

For Windows Guest Files Restore - yes. But you can use Other OS restore wizard as a workaround.

[kevinnaughton]

For Windows Guest Files Restore - yes. But you can use Other OS restore wizard as a workaround.

Perfect. Thanks!