Feature Request v9.5: Can we get Reverse Replication back?

[stjfcic]

At least just for Planned Failovers. It makes running extended tests of the DR site less of a headache. I remember this being a feature in v8 but for some reason it seems to have taken a hike.

[Shestakov]

Hello and welcome to the community!
Do you mean Replication Failback?
Thanks!

[stjfcic]

Thanks!

No, failback implies that I'm moving production back from my DR site (Site B for simplicity) to my Main site (Site A). No I'm just talking strictly reversing the flow of replication jobs after a failover from Site B to Site A.

[foggy]

I'm not sure Veeam B&R ever did this. What it did and still does is reconfiguring replication job to add the original VM to exclusions. But I can't recall it created a reversed replication job, this had to be done manually.

[stjfcic]

Maybe I mis-read something (Edit: I DEFINITELY mis-read something. Sorry about that, it wasn't ever a feature, my bad.), but that's neither here nor there with regards to the request. Currently we are replicating every 30 minutes (15 in some cases) to minimize data loss.

If we need to test the functionality of our DR site, my bosses would like peace of mind and want that test to run a full month (at least). As part of that we need to then replicate our machines from Site B to Site A for that month just in case something were to happen at Site B.

As it sits right now, unless I've missed something, in order to do that I would have to:

1. Have replication jobs running from Site A to Site B
2. Run a planned failover for all VMs from Site A to Site B
3. Permanently failover all VMs to Site B
4. Delete all VMs on Site A
5. Create NEW replication jobs to replicate the servers from Site B to Site A
6. Delete all old replication jobs as they are now obsolete

And when we fail back in a month I would have to do this whole procedure all over again.

We have the bandwidth and infrastructure to support this kind of a procedure but having Veeam shortcut the whole procedure since we're talking about servers that should be in the same state at failover (or at least very close) would be very handy rather than me having to go through and manually rebuild 19 jobs.

[foggy]

I understand your scenario. As a workaround, to minimize amount of work, you could have two sets of jobs in both directions and use them depending on the location where VMs are currently running.

[skrause]

I am in the middle of setting up the Step 5 for failback of our production VMs and I used replica seeding to use the original VMs as the seed. So far in my initial testing it seems to be working fine and should alleviate the need to for step 4. (We have issues with being able to recover deleted thin-provisioned blocks on our storage so the fewer mass deletions I have to do the better).

Granted, we don't use Veeam replication for anything with a high change rate so YMMV.

[stjfcic]

Yeah, I was going to try using replica mapping for the reverse jobs. Just bothers me that now I've got a list of 38 jobs instead of 19.

[skrause]

I always replicate from a Veeam server at the target location so that in a DR scenario I am not relying on a B&R configuration that is in the location that is not in ideal operating shape. Of course, a "reverse this process" being built into the same jobs would be nifty.

[johnlong]

So basically a move option. Where you would move VM's in a replica job to site b and veeam automatically reverses replication to site A using the VM's already there saving the time in having to do a full replication or having to recreate the jobs in reverse.

I use a different software to replicate my VM's and it has this option. It's really nice. Maybe when Veeam implements true CDP in version 10 this will be possible. (Fingers crossed)

[smurugun]

I understand your scenario. As a workaround, to minimize amount of work, you could have two sets of jobs in both directions and use them depending on the location where VMs are currently running.

Hi Foggy,

What about the Veeam license requirement for your above comment. DR site also need to be licensed with Veeam rite?

[foggy]

Hi Murugan, provided only a single set of jobs is running at any particular moment in time, you should be fine to cover this by a single license (you would need to revoke the license and assign it to the DR site servers, this doesn't differ from the original case with failover).