Proxy auto selection according to location on ESXi

[VladV]

According the release notes from the latest patches for VBR the EnableSameHostHotaddMode enables the Backup server to select the proxy according to it's location on the same Host as the source VM for hot add mode.

It seems to work ok when the proxy residing on the same host is up but take this for example:

In an environment with 2 hosts and 3 proxies (one for each host and the third for NBD backup which is Physical), if one hot add proxy is down, VBR will choose the next hot add proxy (which isn't on the same host as the source VM) instead of choosing the NBD one.

Is this the correct behavior or am I doing something wrong? I was thinking that VBR should not choose any other hot add proxy if it resides on another host than the source vm.

[foggy]

Vlad, the behavior is correct. With the default proxy selection algorithm (i.e. when the key is not specified), direct SAN and hotadd proxies are preferred over network mode proxies. NBD proxy is never selected if there is at least one available direct SAN or hotadd proxy. The same logic applies in cases where the key is specified but there're no available backup hotadd proxies on the same host.

[VladV]

It would be really helpful if Veeam will be able to mimic the behavior I mentioned above. In my case, for example, I cannot afford to have our exchange server locked when committing the snapshot because someone forgot to power on the proxy or move it back to the host after maintenance.

I do not know how hard it is to have this functionality, but please consider it as a feature request if possible.

Thanks.

PS: Or maybe help me in finding another method of speeding up the backup for VM residing on NFS storage . I already have 10g links end to end but I can only saturate the storage by having more than 2 disks backed up at the same time (I don't know why). Sometimes though, this is not possible for file servers which have one large disk. Currently our backup speed for single disk VMs is around 140MB/s, but when running multiple disks/VMs backup at the same time (even though they are on the same storage) we speeds in excess of 200MB/s

[foggy]

Yes, I see the case, thanks for the feedback.

Currently, you need to somehow make sure there's always at least one hotadd proxy on the host where the VM resides.

[dellock6]

A simple workaround to this is to edit the backup job and instead of using "automatic selection" you simply select the proxies you want to be used for the job. In this case you can configure the job to use the local hotadd proxy and the nbd one, in this way the remote hotadd will never be used, as long as the VM is not moved to the other host.

[VladV]

as long as the VM is not moved to the other host.

Unfortunately for us, this is a pretty big limitation, but nevertheless I will see if we can find a solution that revolves around sticking VMs to specific hosts until a better solution arrives.

In my opinion though, if Veeam manages to add this functionality to the already good resource management system, it will really bring NFS backup performance closer to iSCSI/FC san mode performance. For example, using 4 proxies, 2 per host, and running all our job simultaneously, we managed to reduce our full backup window with 30% comparing to 10g end to end NBD backup, jumping from 160MB/s to over 300MB/s aggregated speed.

[VladV]

Does this key work with version 8 of VBR or was it included in the GUI?

[Gostev]

Yes, it works in v8, and is still remains the registry key.