How can a IP geolocated to Los Angeles be used as a low-latency EUW endpoint?
I'm trying to understand a routing scenario I've encountered with a game server.
An address, 162.249.72.5 (AS6507, within 162.249.72.0/21), is used as the UDP endpoint for an EUW game.
Several IP geolocation databases place this address in Los Angeles, California. However, traffic from Europe reaches it with latency consistent with a European endpoint, so the packets clearly aren't making a normal round trip to California.
I'm familiar with BGP at a basic level, but I'm having trouble understanding what the underlying architecture would be.
What mechanisms can make a public unicast-looking IP behave this way?
For example, would this normally involve:
- the prefix being announced from multiple geographic locations;
- anycast;
- more-specific BGP announcements;
- ISP-specific routing/peering;
- or some mechanism after the traffic enters the provider's network?
I'm not assuming that anycast is being used in this particular case. I'm trying to understand what mechanisms are capable of producing this behaviour.
I can see the traffic in Wireshark and have confirmed the destination IP, but I'm deliberately not posting my complete traceroute because the visible private 10.x.x.x hops could reveal information about my ISP's internal topology.
I'm mainly interested in the BGP/routing architecture that can make this possible.