Well, looks like cox is having congestion issues again today. Just 2 days before, it was with TiNet handoff in Dallas, and GBLX hand off in LA - DALLAS
Either congestion or an improperly configured aggregated link bond causing unbalanced traffic flow as to another target IP within the same AS, latencies are fine both ways.
Looks like congestion is occuring when it hits cox's edge with Level3 on certain ports.
Doing some very basic troubleshooting here by figuring out which ports have a unexpected increase in latency, and not resolving the IP's host with MTR. Here's the list so far. These tests were done to a COX IP, not simply just pinging the link. You can try this for yourself by going to various prefixes in LA and just watching it take different paths, and certain paths having a large increase in latency.
Congested:
4.53.230.42
4.53.230.70
4.53.230.74
4.53.230.94
Non-congested:
4.53.230.86
4.53.230.66
4.53.230.82
4.53.230.90
One would assume even though COX purchases transit from Level 3, they would make sure their ports weren't over subscribed or saturated.
Non-authoritative answer:
Name: COX-COMMUNI.edge2.LosAngeles9.Level3.net
Addresses:
4.53.230.86
4.53.230.42
4.53.230.66
4.53.228.2
4.53.230.2
4.53.230.82
4.53.230.94
4.53.230.38
4.53.230.90
4.53.230.70
4.53.230.74
4.53.230.34
Heh doing these testing also made me discover that indeed some users are able to get 4-6MS from their gateway instead of the typical 6-8MS. Taking the exact same paths. Yay for those that enable pings to their external IP's
