Amazon Web Services servers don't return ping requests due to them blocking ICMP packets to prevent DDoS attacks, so that testing is unreliable. COX denies that this is their problem because there is no packet loss up to the point of the Phoenix server, 68.1.1.14. The packet loss occurs somewhere after that point. But this happens to so many COX customers in the Southwest, and has been happening to me for nearly a year. From my limited knowledge, I have to say that the "routing tables," or whatever controls the path our connections take to these AWS servers, is completely messed up.
Yes, I wouldn't be at all surprised that ICMP requests are blocked from the AWS Servers for the very reason you mentioned. However, that doesn't negate the validity of isolating the point of network congestion which likely occurs en route to the servers (i.e. Level 3 backbone). Hence, my suggestion for performing a traceroute, then perform pings to the visible hops seems feasible in trying to get a handle on where the packet loss seems to be occurring. Would you not agree?
