Latency

Latency is the time it takes data to travel between two points on a network, usually measured in milliseconds as round-trip time (RTT). The floor is set by physics: light in optical fiber travels at about 200,000 km/s, giving roughly 1 ms of round-trip latency per 100 km of fiber route, before adding equipment, routing, and queuing delays. Typical figures: under 1-2 ms within a metro area, 60-80 ms across the Atlantic, 150-250 ms from Southeast Asia to Europe. Latency drives data center site selection for trading, gaming, real-time communications, and interactive AI inference, and motivates edge deployments close to users. Within AI clusters, east-west latency between GPUs — measured in microseconds over InfiniBand, RoCE, or NVLink — constrains training efficiency at scale.