1 Etymology and definition
1.1 Word origin
The word latency comes from the Latin *latere*, meaning “to lie hidden” or “to be concealed.” This origin reflects the idea of something present but not yet apparent in its effects. In English usage, the term developed to describe a delay between a cause and a visible result.
1.2 General meaning
Latency is the interval between an initiating event and the first observable response. It may describe a signal, request, command, or action and the time before it produces an effect. In many fields, the term emphasizes delay rather than total task duration.
1.3 Distinction from related terms
Latency differs from throughput, which measures how much work a system can complete over time. It is also distinct from duration, which refers to the full length of an event rather than the waiting time before response. In technical contexts, latency often refers specifically to responsiveness.
2 Measurement and units
2.1 Common units of latency
Latency is commonly measured in milliseconds, microseconds, or seconds, depending on the system being studied. Very fast electronic or computing processes may require microsecond or nanosecond scales, while human or organizational delays are often described in seconds or longer.
2.2 Measurement methods
Latency can be measured from the moment an input is sent until a defined output appears. The exact method depends on the system, but it usually involves timestamps, signal monitoring, or direct observation. Accurate measurement requires a clear definition of the start and end points.
2.3 Average, minimum, and maximum latency
Reported latency may be expressed as an average, minimum, or maximum value. An average gives a general picture of performance, while minimum and maximum figures show the best and worst observed cases. These different measures can reveal variability that a single number would hide.
2.4 Latency distributions
Because delay often varies from event to event, latency is frequently described as a distribution rather than a fixed value. Distributions show how often different delays occur and can indicate instability or inconsistency in a system. In practice, percentiles are often used to summarize these patterns.
3 Types of latency
3.1 Network latency
Network latency is the delay involved in sending data across a communication network. It includes travel time, processing time, and waiting time within intermediate devices. This form of latency is central to internet performance and remote communication.
3.2 Computer system latency
Computer system latency refers to the delay between a command and the resulting action inside a machine. It can occur in processors, memory subsystems, storage devices, or software layers. The term is used broadly to describe responsiveness in digital systems.
3.3 Audio latency
Audio latency is the time gap between generating a sound signal and hearing it reproduced. It matters in recording, live performance, and interactive audio software. Excessive delay can make monitoring and synchronization difficult.
3.4 Video latency
Video latency is the delay between capturing an image and displaying it. It is especially relevant in streaming, video conferencing, and machine vision. Even small delays may affect interactive applications or coordinated actions.
3.5 Mechanical latency
Mechanical latency is the time required for a physical device or mechanism to begin or complete a response after activation. It can arise from inertia, friction, control systems, or actuator limitations. This type of latency is common in engines, robotics, and industrial machinery.
3.6 Human response latency
Human response latency is the delay between a stimulus and a person's reaction. It is studied in psychology, physiology, and user-interface design. Reaction time is a common example of this kind of delay.
4 Latency in computing
4.1 Input and output delay
In computing, input and output delay describes the time between a user action and the system’s visible response. This includes keyboard, mouse, touchscreen, display, and peripheral delays. Low input delay is important for interactive applications.
4.2 Storage latency
Storage latency is the wait time before data can be read from or written to a storage medium. It varies among hard drives, solid-state drives, and networked storage systems. Access time and controller overhead both contribute to this delay.
4.3 Memory latency
Memory latency is the interval between requesting data from memory and receiving it. It affects processor efficiency because the CPU may have to wait for needed information. Faster memory access can improve overall system responsiveness.
4.4 Processing latency
Processing latency is the time required for a computer to carry out a task after receiving input. It may depend on the complexity of the operation, the speed of the processor, and software efficiency. In real-time systems, this delay can be a critical constraint.
4.5 Latency in operating systems
Operating systems influence latency through scheduling, interrupt handling, and task management. They decide when processes run and how resources are shared among them. Poor scheduling can increase delay even when hardware is capable of faster response.
5 Latency in networking
5.1 Propagation delay
Propagation delay is the time a signal takes to travel through a medium. It is affected by physical distance and the speed at which signals move in cables, fiber, or wireless channels. This component cannot be eliminated, only reduced by shorter paths or faster media.
5.2 Transmission delay
Transmission delay is the time needed to place all bits of a message onto a communication link. It depends on message size and link capacity. Larger data packets generally require more transmission time.
5.3 Queuing delay
Queuing delay occurs when data waits in line before being processed or forwarded. It becomes more noticeable when network devices are busy or traffic is heavy. This delay can fluctuate significantly under changing load conditions.
5.4 Packet loss and jitter
Packet loss and jitter can worsen the effects of latency in communication systems. Packet loss may require retransmission, which adds delay, while jitter refers to variation in arrival time. Both can disrupt smooth delivery of real-time data.
5.5 Round-trip time
Round-trip time is the total time for a signal to travel to a destination and return with a response. It is often used as a practical measure of network delay. In many applications, round-trip time is more meaningful than one-way latency.
6 Latency in media and communication
6.1 Streaming latency
Streaming latency is the delay between a live event and its appearance on a viewer’s screen. It results from capture, encoding, transmission, buffering, and playback. Lower latency is often preferred for live sports, auctions, and interactive events.
6.2 Real-time communication
Real-time communication systems aim to keep delays small enough for natural interaction. Voice calls, video chats, and collaborative tools depend on prompt exchange of data. Even modest latency can make conversation feel less immediate.
6.3 Broadcast delay
Broadcast delay is an intentional pause inserted before transmission or playback. It may be used to allow moderation, synchronization, or technical correction. This delay can help manage live content without affecting the source event itself.
6.4 Interactive systems
Interactive systems require quick feedback so that users can act and adjust naturally. Delays in these systems may reduce usability and make controls feel sluggish. Designers often aim to keep response times below perceptible thresholds.
7 Latency in science and medicine
7.1 Neural latency
Neural latency is the interval between a stimulus and the corresponding response in the nervous system. It is studied to understand sensory processing, reflexes, and brain function. Measurements may reveal how quickly signals travel and are interpreted.
7.2 Reaction time studies
Reaction time studies examine how long it takes a person to respond to a stimulus. These studies are used in psychology, ergonomics, and clinical assessment. They can provide information about attention, processing speed, and motor control.
7.3 Latency periods in biology
In biology, a latency period is a phase during which a process is present but not yet visibly active. The term appears in development, infection, and other physiological contexts. It generally indicates a hidden or inactive interval before later expression.
7.4 Latent intervals in disease processes
Some diseases involve a latent interval between exposure and symptoms. During this time, a condition may develop internally without obvious signs. The length of the interval can vary widely depending on the illness and the host.
8 Causes and influencing factors
8.1 Distance and physical medium
Greater distance usually increases delay because signals need more time to travel. The type of medium also matters, since copper, fiber-optic cable, and wireless channels have different characteristics. Physical constraints place fundamental limits on speed.
8.2 Hardware limitations
Device speed, component quality, and internal architecture all affect latency. Slow processors, limited memory bandwidth, or mechanical parts can create bottlenecks. Older or lower-performance hardware often exhibits longer delays.
8.3 Software overhead
Software overhead includes the extra time spent managing tasks, translating instructions, or handling abstractions. Operating systems, drivers, and application code may all contribute. Inefficient software can increase latency even on capable hardware.
8.4 Congestion and workload
Heavy traffic or high demand can slow down a system by increasing wait times. This applies to computer networks, servers, and physical queues alike. Latency often rises sharply when resources are close to capacity.
8.5 Buffering and synchronization
Buffers hold data temporarily to smooth variations in arrival or processing speed. While useful, they can add delay before output begins. Synchronization mechanisms may also introduce waiting when multiple components must align their actions.
9 Latency reduction and optimization
9.1 Hardware improvements
Using faster processors, lower-latency memory, and more efficient devices can reduce response times. Improved network infrastructure and shorter signal paths also help. Hardware upgrades are often the most direct way to lower delay.
9.2 Protocol and algorithm design
Efficient protocols and algorithms can reduce overhead and minimize unnecessary waiting. Designs that limit handshakes, retries, or redundant steps often perform better. Careful engineering can make systems more responsive without changing the underlying hardware.
9.3 Caching and prefetching
Caching stores frequently used data close to where it will be needed, reducing access delay. Prefetching attempts to load data in advance based on expected demand. Both methods can improve perceived speed when applied appropriately.
9.4 Parallelization
Parallelization divides work among multiple components so tasks can proceed at the same time. This can shorten response times when dependencies are limited. However, coordination costs may reduce the benefit if the task is not well suited to parallel execution.
9.5 Trade-offs with reliability and quality
Reducing latency sometimes comes at the cost of reliability, accuracy, or quality. Systems may skip checks, use smaller buffers, or compress data more aggressively to save time. Designers often balance speed against stability and precision.
10 Applications and examples
10.1 Consumer electronics
Latency affects the responsiveness of phones, tablets, televisions, and smart devices. Users notice it when tapping a screen, pressing a button, or watching media playback. Manufacturers often highlight low delay as a feature of better performance.
10.2 Online gaming
In online gaming, latency influences how quickly player actions appear in the game world. High delay can make controls feel unresponsive and can affect competitive play. Game developers use networking techniques to reduce the impact of lag.
10.3 Telecommunications
Telecommunications systems rely on low latency for clear voice calls, messaging, and data transfer. Delays can interfere with conversation flow or real-time coordination. Network design and routing choices play a major role in performance.
10.4 Scientific instrumentation
Scientific instruments often require precise timing and minimal delay to capture accurate measurements. Examples include sensors, telescopes, laboratory analyzers, and control systems. Latency can affect both data quality and the timing of experiments.
10.5 Industrial control systems
Industrial control systems depend on predictable latency to monitor and regulate machinery. Delayed feedback can reduce precision or create safety concerns. For this reason, many control environments are designed for consistent, low-delay operation.