Latency denotes the time interval that elapses between an initiating event and the appearance of its effect in a system. In engineering contexts the term describes delays that affect responsiveness and timing; in computing and telecommunications it commonly refers to the delay before data or a response is observed. See general engineering uses at engineering resources for deeper technical context.
Characteristics and components
Latency is not a single quantity but a sum of contributing delays. Typical components include:
- Propagation delay — the time required for a signal to travel through a medium.
- Transmission/serialization delay — the time to place bits onto a link or bus.
- Processing delay — time spent by devices or software handling the data.
- Queuing delay — time spent waiting in buffers or queues under load.
History and terminology
The word "latency" derives from general usage meaning delay or hidden interval and was adopted across disciplines as systems became analyzed in terms of timing and responsiveness. Over decades the concept gained specific technical definitions in networking, audio engineering, control systems and storage I/O as engineers described the distinct delays that affect system behavior.
Uses, examples, and importance
Latency matters wherever timely responses are required. Examples include interactive web pages and APIs, online gaming and virtual reality, audio monitoring and live performance, telephony and video conferencing, high-frequency trading, and control loops in industrial equipment. Low latency improves perceived responsiveness; in real-time systems excessive latency can cause errors or safety risks.
Measurement and trade-offs
Engineers measure latency using round-trip and one-way metrics, timestamps, packet traces and specialized tools (for example ping and traceroute analogues). Latency and throughput (bandwidth) are distinct: a connection can have high throughput but still exhibit long delays. Physical limits, such as signal propagation speed and processing complexity, set practical lower bounds.
Mitigation strategies and notable facts
Common approaches to reduce latency include caching and content delivery networks, edge computing, parallelism and pipelining, protocol optimization, prioritization and quality-of-service controls, and hardware acceleration. Trade-offs often occur: techniques that lower latency can increase resource use or reduce consistency. Awareness of the different delay components helps designers choose appropriate mitigations for their application domain.