Latency across the Pacific tends to look pretty bad under the best of conditions, but overall throughput shouldn't be suffering like that. Of course, when I ran the Speedtest.net test (through to Tokyo to see if I could reproduce the problem) it reported 209ms latency:
A quick trace to square.co.jp (the first japanese hostname that sprang to mind) from a server I've got in our datacenter, not but a dozen yards from my desk, shows:
Code: Select all
HOST: REDACTED Loss% Snt Last Avg Best Wrst StDev
1.|-- REDACTED 0.0% 20 1.2 1.3 1.1 2.3 0.3
2.|-- 2.ge-1-1-0.gw.sr.sonic.ne 0.0% 20 0.4 0.6 0.4 1.6 0.5
3.|-- 0.ae0.gw2.sr.sonic.net 0.0% 20 1.5 1.3 0.4 1.5 0.4
4.|-- 0.ge-6-1-6.gw3.equinix-sj 0.0% 20 4.9 6.0 4.6 14.6 2.7
5.|-- 10gigabitethernet2-3.core 0.0% 20 6.2 9.2 5.2 16.8 3.9
6.|-- pacnet.10gigabitethernet2 0.0% 20 4.6 5.0 4.3 10.2 1.3
7.|-- gi9-0-0.cr2.nrt1.asianetc 0.0% 20 118.1 117.9 117.0 118.2 0.3
8.|-- ge-2-1-0-0.gw3.nrt5.asian 0.0% 20 117.2 117.2 117.1 118.3 0.3
9.|-- GMO-0003.gw3.nrt5.asianet 0.0% 20 118.7 118.8 118.7 119.0 0.1
10.|-- c7-e-1-1.interq.or.jp 0.0% 20 117.8 117.9 117.7 119.0 0.3
11.|-- g-pb1-e-0-1.interq.or.jp 0.0% 20 120.4 126.2 118.4 181.0 13.6
12.|-- lb04.virt.lolipop.jp 5.0% 20 118.5 118.5 118.4 118.7 0.1
I tried again with 1400-byte packets and it only added ~3ms to the round-trip times at the last hop, so I have to take Speedtest.net's latency numbers with a grain of salt. That said, it sounds like a real problem is happening somewhere on the route between ladr's connestion and the land of the rising sun, and a trace would be of great help in tracking it down.