Question about the advanced deployment

[TheHolySnail]

Hi guys

I have a question regarding the suggested advanced deployment in the user guide: https://helpcenter.veeam.com/backup/vsp ... anced.html
If I use physical servers with access to SAN and the LUNs on the production storage system and assign them the proxy role, will the actually read the data over FC and then write it to the configured repository (another physical server) via Ethernet? Or is it possible to forward the traffic to the repository via FC aswell? Assuming my repository is another SAN attached server, of course.

Thank you in advance.

[Shestakov]

Hi Jan!
Do you use VBR server as backup proxy? Backup data flow is always Disk > Backup proxy > Network > Backup repository > Disk. How are proxy and repository connected?
Btw, we are talking about Direct SAN transport mode, take a look at the description and requirements.

[foggy]

If you setup backup repository on the proxy server and give it Fiber Channel access to the storage, you could avoid traffic going over Ehthernet.

[TheHolySnail]

Sorry for my late reply. I was enjoying some (well needed) days off.

Hmm.. as I expected, there is no way to have my backup data going over FC only. I was planning to use physical servers as proxies so it would be easier so scale out if needed. This way around, with proxy and repository on the same host, I'll have a mixed deployment if I add more proxies, which is not very much to my liking. But I'll go with your suggestion, proxy and repository on one host.

My only worry are the two source systems (3PAR 10800 pretty much fully specced) and the customers backup window which needs to be as short as possible. This is why I was thinking about seperate, physical proxies. But I guess I will be fine if I use two servers with enough RAM and CPU horsepower as well as SSDs in my repository (looking at Luca's idea on his blog).

[Shestakov]

Physical proxies with Direct SAN mode is considered as the best practice. I would run several jobs for testing and review bottleneck statistics.

[TheHolySnail]

Physical proxies with Direct SAN mode is considered as the best practice.

This is what I am going to do. Two physical servers, each with the proxy and repository role configured.

I would run several jobs for testing and review bottleneck statistics.

I'd love to. Issue being I am answering an RFP which does usually not include such things..

[Shestakov]

In most cases the bottleneck is source which means the user need to read source data and transmit it to proxy faster. But it should work well in your case because of Direct SAN.
Let us know how it works once you have everything configured!