Traceroutes, ICMP and BGP

This website is super cool: it’s a visual deep dive on how traceroute works.

My traceroute implementation uses a protocol called ICMP. ICMP was designed specifically for sending diagnostic information around the Internet, and, helpfully, almost every Internet-connected device speaks it. Interestingly, ICMP packets have a “TTL” (time to live) field. This isn’t actually a “time” as implied by a name — it’s a countdown! Every time a router forwards an ICMP packet along, it’s supposed to decrement the TTL number. When the TTL hits zero, the router should stop forwarding it along and instead send an error message to the packet’s source IP saying that the packet has reached its maximum number of hops.

We can take advantage of this TTL feature! To do a traceroute, we can send a bunch of ICMP packets with increasingly large TTLs. The first packet with a TTL of 1 will error on the first device it reaches, and so on, until we hopefully get an error back from every routing device that touched the packet. These error packets include diagnostic information like the IP address of the device that sent the error, allowing us to trace your packets’ rough path across the Internet.

This led to this interesting dive into BGP and ASNs.