Failback to Production

[Michael_6835]

I'm testing failover at my DR site, I failed over and ReIP the machine successfully. I updated the vm with a few downloads and windows updates to create changes from the source that i would be able to verify.

I then initiate a failback to production and it seems to be taking longer than expected. Understanding that "longer" is all relative, i'll detail my setup.

Veeam Console at DR site - Serves as DR proxy and site where DR replication jobs live.


Production site - Has Proxy connected to san via iscsi initator.

When doing backup and replications, I manually specify source and target proxies.
On production proxy, it uses SAN mode, on target proxy it uses "hotadd"

On the failback to production, it tells me, source proxy (DR) uses hotadd (ok), but target proxy (Production site) uses "NBD".
I suspect the reason for the delay on replication back is the use of NBD.

Would that be correct? And if so, how come it can't use SAN mode for the target proxy failback? I'm trying to understand if I have something wrong with my design or if this is simply how the product was intended to function.
Thanks

[veremin]

I'm trying to understand if I have something wrong with my design or if this is simply how the product was intended to function.

This is by design. Direct-SAN mode can’t be used during failback operations, as the result, proxy fails to NBD. In order to increase failback performance you might want to utilize Hot-Add proxy, instead. Thanks.

[Michael_6835]

Using hot-add proxy would mean i need a virtual machine on the production site defined as a proxy right?

Currently at production, the proxy is a physical server, at DR it's a Virtual machine.

[veremin]

Yep. The role of Hot-Add proxy can be assigned to any Windows-based virtual machine. Thanks.

[foggy]

Besides, during failback Veeam B&R needs to calculate and synchronize the differences between the original VM and the replica VM so the time required to scan the VM image depends on its size and is comparable to the full backup time of this VM. Also, be sure you have specified the repository at the source side to store replica metadata in the replication job as failure to do so could seriously affect failback operation performance.

[Michael_6835]

Adding an additional proxy at the production site that was a virtual machine was the solution. I added this proxy simply to use hot add for the fail back procedures. Fail back took less than an hour.

thanks for pointing in the right direction.
Mike

[veremin]

You’re welcome. Should any other questions arise, don’t hesitate to let us know. Thanks.