Backup Proxy Design Question

[htwnrva]

Implementing v6 from v5 today and I have a question about my design of the VI Proxies. I have 3 physical Veeam servers at 2 locations, each has a bunch of local drives as well as some SAN drives set up as repositories

Server1 - Site 1
Veeam Installation
Backup Repository created for each of the D,E,U,V,W drives

Server2 - Site 1
Backup Repository created for each of the D,E,U,V,W drives

Server3 - Site 2
Backup Repository created for each of the D, E drives

By default when I create a backup job it's set to Automatic selection of proxy but I notice that it's using Server3 to do backups of Site 1 and Server1 and 2 to do backups at Site 2. Obviously don't want that data going over the pipe on a regular basis. So I need to edit each job at Site 1 and remove Server3 from the available proxies. Then I need to edit each job at Site 2 and remove Server1 and Server2.

The question is how much of a performance hit or network usage will incur if Server1 is used to backup to drives attached to Server2 and vice versa. Physically they are racked on top of each other so they will just hop through 1 network switch.

[foggy]

Ronny, you have not specified the location of your hosts. Note that automatic proxy selection mechanism (described in the sticky FAQ in detail) picks up the most preferable proxy for each VM in the job, so it is recommended to place the proxy closer to the source host. Most likely this is the reason of such proxies selection in your environment.

Besides, if you use the Windows or Linux server as a target, then the data come across the network (from proxy to the backup repository) already compressed and deduped and not "on a regular basis".

[htwnrva]

Got it, so it would be OK for any server at site 1 to process the data and send to the drives on the other server because it's already deduped and compressed. The hosts are both at site 1 and 2, but i only want site 1 hosts to be backed up by site 1 servers and vice versa.

[Vitaliy S.]

Got it, so it would be OK for any server at site 1 to process the data and send to the drives on the other server because it's already deduped and compressed.

Correct.