traceroute, mtr, ping: Where Is the Latency?

A remote service can feel slow even when every request eventually succeeds. The first diagnostic question is whether the delay appears consistently at the destination, intermittently along the path, or only in the application exchange.

Three command-line tools provide related but different evidence:

None of these tools measures the complete latency of an API call. Their probes can use different protocols, ports, packet sizes, and routes from application traffic. Routers can also forward packets quickly while delaying or discarding diagnostic responses. Correct interpretation depends on what reached the final destination, not on the largest number in an intermediate row.