It may describe one-way delay from sender to receiver or round-trip time from sender to receiver and back. Meaningful results therefore state the direction, endpoints, packet type, and observation interval rather than reporting a number without context.
Delay can accumulate through propagation, transmission, processing, and queuing. It varies with route, load, link technology, packet size, and traffic handling. Averages can conceal spikes, so operations teams commonly examine percentiles and delay variation alongside representative application measurements.
Key points
Measurement choiceOne-way measurement can reveal directional differences but requires adequately synchronized clocks. Round-trip measurement works from one endpoint without clock synchronization but combines both directions and the remote system’s response behavior.
Operational useCompare like-for-like measurements with a service baseline, correlate changes with path, utilization, loss, and application evidence, and test from locations that represent affected users.
Performance effectHigh or variable latency can slow interactive applications, delay acknowledgments, increase transfer time, and impair real-time voice or video even when link capacity is available.
Important limitationLatency is a performance observation, not proof of congestion, attack, or a particular faulty device. Active probes also add traffic; obtain authorization, rate-limit them, and follow site-specific safety-assessment and change-control requirements on operational technology networks.