Wierd but 100% consistent replication failure

[Natoli]

Okay, I already contacted technical support with this issue and we didn't get anywhere, so maybe some forum lurkers might have ideas for me.

I have two ESXi 5 hosts that were upgraded from ESX 4.0. Before any upgrades all of my replicas ran without issue.
After I upgraded, I brought my offsite backup storage server in house to increase the storage space and reformatted fresh to ESXi 5. While I had it local, I setup a static route (to maintain the WAN IP of the target) and did the initial replications of all my jobs (all of which were also recreated.) The initial jobs went off without a hitch. I removed the static routes and sent my server back offsite.
Now, certain jobs fail every time with what appears to be connection failures.
Here is where it gets interesting.

Some of my jobs run fine regardless of the time ran or how long they take.
As I said, only some jobs fail; and those jobs fail 100% of the time, always at the same processing point per VM.

For example, I have one VM that I'm trying to replicate. The job will launch and start processing. On B&R v5 it processes exactly 2.4GB and fails. I deleted the failed replica from target disk, deleted the job, and created a new job. Same result!
So then v6 was released. I installed the trial version completely separate from v5. Deleted all traces of the replica from the target. Created a new job. In v6 processes 2.5GB and fails. Which I'm under the belief it is actually the same as the 2.4GB from v5, just different calculation/rounding.

So then I proceeded to watch the target Events and Tasks windows at the failure point. What's happening is rapid login/logouts and remote disk opened for read/write each successive login.

My feeling is that something is wrong with the VM host, not the target or the connection.
Anyone ever seen any behavior like this? Or suggestions on what you would try next?

[Natoli]

Thought I'd post back regarding this.
Evidently it is a network related problem, albeit a very strange one.

I went created a Veeam Proxy on my WAN side and my jobs run now.