Sure the ping is stable all day long at about 26ms. Since the remote is usually VDSL (they just use what is the cheapest for them in order to get to the bandwidth the customer wants) that seems alright. At the remote sites we put Cisco SG300s and at the DC we have Enterasys Core switches, so that shouldn't be a problem to get a 10Mbps link going.

Funny thing that I found out, a single TCP transfer starts to stall, however using multiple parallel transfers we get to the bandwidth.

Oh and by the way, I have of course also tested using a mostly "direct" connection (laptop connected without switch directly at the remote site, and counterpart at the DC on the same switch than the incoming connection there)

@marki I assume that there was no packet loss when you pinged through the new metroe service, correct? I suspect that you will need to implement some qos policies... if you have to spend a few thousand USD to fix the problem, is that an option?