Latency

Delay between two points, measured one-way or round-trip, set by distance, hops and queues, and measured along a path rather than a link.

Definitions only — a term explained here is not a service we sell.

Latency is the delay between two points, shaped by distance, hops and queues. It is measured one-way or round-trip, and each figure should say which. Bandwidth is the pipe, not the trip, so a fast link can still feel slow.

What the number includes

Propagation sets the floor: light in fiber travels about 200 km per millisecond, so a 10,000 km path has a one-way floor near 50 ms. Queueing is the variable part: a link at 90% utilization adds delay and jitter.

Where the delay comes from

40 ms

Where the delay comes from
Value What it means
1 ms Same metro
10 ms Regional path
40 ms Regional path
120 ms Intercontinental path
300 ms Queues or a long path
800 ms Queues or a long path
Round-trip time on a scale, from same-metro paths to queues and long paths.
Text description

The ruler plots round-trip time on a scale. Short metro paths sit low, regional and intercontinental paths in the middle, and queues or long paths high.

What to check

An average hides the tail, so ask for percentiles, the measurement method and the measurement points. One figure describes one path at one time. ICMP ping measures the network, not the application: each TCP handshake, TLS negotiation and database query adds round trips. Measure from the user's network at peak with the Fiber latency calculator.

FAQ

What is a good latency?

It depends on the application. Voice guidance treats 150 ms one-way as the upper bound for a natural conversation; a round-trip figure is about twice that.

What is jitter?

Variation in delay between packets. Real-time voice and video suffer even when the average latency is fine.

Does a longer route always mean higher latency?

Usually. A congested short path can be slower than a longer quiet one.

Learn more

Sources

  1. ITU-T G.114: One-way transmission time