Two Lines from DC to DR

[digitalexpl0it]

Hello,

I am trying to figure out how to make replications go out the correct line. I have two lines from my DC to my DR. One is a 10 MB MPLS the other is a 1 GB fiber. My issue is that everything is going through the MPLS and not the 1 GB fiber. When I do a traceroute from the veeam server in DC to a veeam proxy in DR it goes through the 1 gb line just fine. I have a case open but not much help, hard to get someone and if you do get someone on the call they are pulled away because they are taking to much time and the supervisor makes them drop the call... GREAT SUPPORT by the way. Also beef is this 7PM only help or help via email with articles... humm I could search for those articles oh wait I already did this might be why I am contacting Support? ....$40,000 upgrade sure doesn't get you much...

I have a veeam proxy for each backup job and replication job on both the DC and the DR.I have tried WAN acceleration with the same issue. I am seeing 1MB-10MB transfer rates when I should be getting over 150MB. This all started after I upgraded the license to enterprise plus.. Perhaps this was a mistake. Having to go on the manufactures forum for help is very bad. I am not even sure help will be here. I have many issues with the Veeam software but I must put in a ticket for each issue and then I end up with multiple techs that just want logs.

I have setup my replica jobs to use network mapping, i have removed the second NIC (for production use to access the proxies) and only have the one NIC on the backup network. I have set routes manually on the devices. Still nothing

Open Cases:
01858345
01858344
01856074

So for now, I have backups that are failing, replicas that are failing, replicas that are so slow that it is taking days and they are not finishing and I was just emailed that its not critical 1 and will be moved down from Veeam support.. humm so no replicas or backups is not critical? So basically the Tech has giving up and I am loosing lots of faith in this product staff and how things are ran.

[csinetops]

How are you routing traffic to the two lines? Couldn't you route traffic from your backup network to the fiber connection? For instance, we have a MPLS down to our DC, we have routes set up on our core switch that says if you are looking for the 192.168.x.x network go over MPLS, else dump to edge gateway. Your 10MB and 1G have to have routes set up somewhere for the traffic to know where to go. Set a route for your backup traffic to flow over the 1G instead of the 10MB.

[zoltank]

This sounds like a routing issue and not a Veeam issue.

[foggy]

Derrick, have you configured preferred networks for data transfer?

[digitalexpl0it]

How are you routing traffic to the two lines? Couldn't you route traffic from your backup network to the fiber connection? For instance, we have a MPLS down to our DC, we have routes set up on our core switch that says if you are looking for the 192.168.x.x network go over MPLS, else dump to edge gateway. Your 10MB and 1G have to have routes set up somewhere for the traffic to know where to go. Set a route for your backup traffic to flow over the 1G instead of the 10MB.

Correct we do the same, So basically anything on the 10.2.x.x or 10.3.x.x goes out the 1 gb fiber while anything with 172.16.x.x or 192.168.x.x goes through the MPLS, When I do a trace route it routes correctly on all the proxies Windows and Linux

[digitalexpl0it]

Derrick, have you configured preferred networks for data transfer?

Yes Veeam support went though this and we tried it, didn't have any affect on how the traffic was routed.

[digitalexpl0it]

This sounds like a routing issue and not a Veeam issue.

Could be, I went though all the veeam proxies and they did have two nics on on the backup subnet 10.x and one on the prod network, I removed the nics completely to the prod network. I did see a speed up on local backups but still replicas are going out the wrong pipe

[zoltank]

I don't know if it's doable, but could you remove any 172 or 192 addresses on the Veeam server, run a replication job, and see if that makes it go through the 1gb fiber?

[Jamie]

Doubtful it is Veeam, sounds like DNS and routing.

Veeam is probably doing a DNS lookup and finding the server on its 'normal' network, and going out that interface / gateway to get to it.
Try creating a Host file entry on server1 that references the server2 under server2's "fiber network" IP only.
When Veeam on server1 tries to look for server2, it will hit the host file reference rather than doing a DNS and go out the correct interface.
You may have to add a route for that network in the OS, depending on how you network is setup.

[unsichtbarre]

I think the fact that the traffic routes from proxy to proxy, not "Veeam Server" (which is also a proxy) to proxy is being left out of the discussion. digitalexpl0it, you said that there was one Veeam Proxy per job.

If you tracert from your Veeam Server (at DC) to your Veeam Proxy (at DR) the result does not indicate the actual path of network traffic. You need to tracert from Veeam Proxy (DC) to Veeam Proxy (DR)

Also, and I have done this alot, one proxy per job probably does not increase performance. We'll increase to 8 or 12 vCPU on one Veeam Proxy (with concurrent processing), prior to adding another one.

[digitalexpl0it]

So the routing is now going through the 1 gb. I did remove all the 172 and 192 ips from the proxies and veeam server. Ran pathping, traceroute and used iperf. Looking at my netflow monitor I see the traffic going the correct direction. Whats weird is I sue IP's for everything so DNS shouldn't be used.