Force Veeam to use 10 gigabit network

[nmace]

I've got Veeam installed on a physical server. I've got a new 5.5 cluster running on top of Nutanix nodes on a pair of 10 gig switches, the switches have working 1 gig uplinks to the core switch....which the physical Veeam server is also plugged into.

I've connected the Veeam server to the 10 gig switches using a 10 gig NIC on the Veeam server, it has a separate IP although it's the same sub-net/vlan as the "normal" 1 gig connection on the Veeam server. How do I tell Veeam to backup the VMs via the 10 gig connection instead of going around via the core switch + uplink ports? I'm thinking it involves adding a backup proxy, but I'm not sure as it would also be able to talk to Veeam via the 1 gig uplink port as well. How do I get around this issue? This wasn't an issue with my old cluster with a SAN, as the 10 gig network for that cluster was isolated and a different sub-net, so the only way Veeam could get to the data-store was via the 10 gig connection.

[Gostev]

Just register your backup proxy and backup repository using IP addresses from 10Gb NIC, and the traffic will go over that network. Veeam server will also need a NIC in 10Gb network to be able to communicate with those, of course.

[nmace]

Just register your backup proxy and backup repository using IP addresses from 10Gb NIC, and the traffic will go over that network. Veeam server will also need a NIC in 10Gb network to be able to communicate with those, of course.

I'm not sure what you mean by register my backup proxy and backup repository using IP addresses.

I created a new VM on the new cluster, made it a backup proxy, and selected it as the proxy for the job to use, and when the job is running I have 0 bits going out my 10 gig NICs on the Veeam server. Everything is going out the 1 gig NIC. Any ideas?

Edit: the job is in fact using my newly created proxy VM, per the job details. It just is getting to it via the 1 gig network instead of the 10 gig network. Where it's all the same subnet, I'm not sure what I am trying to do is even possible.....

[dellock6]

Anton means to add the proxy VM in the Veeam console by IP.
What probably is happening is like in this example:
- you have a windows vm with name proxy.domain.local
- say it has 192.168.0.10 in the 1G network and 10.0.0.10 in the 10G network, you configured active directory dns in the primary nic because that one is the network where the domain runs
- you joined an AD domain and because of your configuration, now dns has a record where proxy.domain.local resolves to 192.168.0.10

If you added the vm to veeam console using its name, every operation involving that proxy flows via the 1G network as you are seeing.

Try to remove the VM from Veeam, and add it again by using 10.0.0.10. In this way all activities of the proxy will flow via the 10G network (of course, if also Veeam console and repository can run on that network).

Luca.

[nmace]

Anton means to add the proxy VM in the Veeam console by IP.
What probably is happening is like in this example:
- you have a windows vm with name proxy.domain.local
- say it has 192.168.0.10 in the 1G network and 10.0.0.10 in the 10G network, you configured active directory dns in the primary nic because that one is the network where the domain runs
- you joined an AD domain and because of your configuration, now dns has a record where proxy.domain.local resolves to 192.168.0.10

If you added the vm to veeam console using its name, every operation involving that proxy flows via the 1G network as you are seeing.

Try to remove the VM from Veeam, and add it again by using 10.0.0.10. In this way all activities of the proxy will flow via the 10G network (of course, if also Veeam console and repository can run on that network).

Luca.

I see what you mean now, but that isn't the issue I'm having. Now that I've had some sleep and coffee, let's see if I can do a better job of explaining what's going on.

My physical veeam server is 10.100.5.x. My VMs on the new cluster are also 10.100.5.x. The hosts are all connected to a pair of 10 gig switches that's used for NFS/vMotion/VM Traffic. Those switches have an uplink to the core switch for the VM traffic that needs to traffic outside of the cluster.

I've added a 2 port 10 gig NIC to the Veeam server, plugged it into the switches and assigned an IP to the teamed NICs (also on the 10.100.5.x network). However when I tell Veeam to backup one of the VMs in the cluster, it does so by hitting the core switch, traveling up the uplink ports on the 10 gig switches and backing up the VM that way instead of just using its own 10 gig NICs that go straight to the 10 gig switches. I added a VM to the cluster and installed Veeam's software on it to make it Veeam proxy and edited the job so it uses that proxy. It does in fact use the proxy, just via the 1 gig network path instead of the 10 gig path...which is the root issue I'm trying to resolve.

I believe the problem is that everything is on the same subnet/vlan, so it's using primary interface to access it (primary being the one interface with a default gateway, Windows complains if you assign more than 1 gateway). I guess what would be ideal is if there was a setting in the Veeam job settings to force a job to use a specific NIC. That doesn't seem to exist....so what do I do?

[dellock6]

Ah, got it!
So, you have two quick solutions: you need to change the default network adapter (http://technet.microsoft.com/en-us/libr ... s.10).aspx) or even better, why don't you disable the 1G network, and move the old IP address to the 10G trunk? Otherwise, since they are on the same network the default nic will always be used...

Luca.

[nmace]

Ah, got it!
So, you have two quick solutions: you need to change the default network adapter (http://technet.microsoft.com/en-us/libr ... s.10).aspx) or even better, why don't you disable the 1G network, and move the old IP address to the 10G trunk? Otherwise, since they are on the same network the default nic will always be used...

Luca.

There are some other things that server is used for, specifically it has a raid volume shared that's a landing spot for some database backup scripts. If I go with the second option, that traffic would have to transverse the uplink ports of my cluster switches....which I don't want. I'll check out that link, thanks!

[nmace]

That link seems to be for Server 2008, 2008 R2, and Windows 7. My Veeam server is on Server 2012, and that option doesn't seem to exist. At least not as described. And Googling for that seems like it would simple to find, but I'm not having any luck. Anyone know how to do this on Server 2012?