Network Latency
Also known as: Latency
Definition
Network latency is the time interval between sending a packet or frame and receiving it at the destination, often measured as one-way delay or round-trip time. It reflects propagation, serialization, queuing, and processing delays along the path.
Key Points
- Latency is measured as delay across a network path.
- It is shaped by distance, routing, queuing, and processing overhead.
- Satellite links usually have much higher latency than terrestrial fiber or metro networks.
- Latency is a performance metric, but it also affects service assurance and operational control.
- Jitter and packet loss often make latency more harmful than the raw number alone suggests.
Concept
In practice, latency is a composite result of physical distance and network behavior. A long path adds propagation delay, but congestion, routing detours, and processing can add just as much impact in busy or inefficient networks. Operators therefore look at latency alongside jitter and loss rather than treating it as a standalone number.
Latency is especially important for interactive, time-sensitive, or control-oriented applications. When delay crosses a threshold, voice, collaboration, remote control, and automation workloads can become unreliable even if the network is otherwise “up.”
Explainer
The main limitation of latency is that it is not only about geography. A short path can still be slow if queuing is heavy or if the architecture adds unnecessary hops, tunnels, or policy detours. That means low latency depends as much on design and congestion management as on physical distance.
A second issue is that measured latency can hide variability. Stable average delay may still include spikes that break interactive applications or control loops. This is why latency is usually analyzed with jitter, loss, and application tolerance rather than in isolation.
Across ConnectedEarth sectors, latency matters in satellite communications, SD-WAN, industrial control, remote operations, aviation, maritime, and cloud-connected enterprise networks. In each case it influences what kinds of applications can be supported and how resilient the user experience will be under load.