If you have access to a client that can reproduce high end-to-end latencies, sending us a traceroute (ideally using, mtr -bzo "LSD NABWV" -r -n [ip] for most helpful formatting/info) could help us identify any sub-optimal routing and potentially improve things.