Well, at the moment i can't pin it down to EC2, but it's the only thing i can imagine it'd be. The network/cpu/memory usage is all healthy, and there's nothing in the kernel log, so that's what i'm guessing is the cause. Although i may be wrong.
We've observed the bottleneck to be an upper limit on packets/sec for a given instance type. On an m1.large this is about 100k/sec. I believe it's due to the virtual NIC just not being fast enough to handle high traffic loads.