Definition
Latency is the time between sending data or a signal and the moment it arrives, or the moment you get a reply. It's the key measure of how instant things feel in internet connections, games, video calls, and more.
How Is Latency Different from Bandwidth?
Even if you mail a letter in a huge truck, the trip to its destination still takes a set amount of time. Internet connections work the same way. How much data you can send at once and how fast a signal actually travels are two different things. If bandwidth is the number of lanes on a highway, latency is the actual travel time a car needs to get there.
A 10-lane road with a 20 mph speed limit still gets you there late. A single lane where cars run at 60 mph delivers one small package much sooner. When you download a huge movie, wide lanes (bandwidth) help. But for something fast-reacting like an online game, an environment with low latency matters far more.
Latency is usually measured in milliseconds (ms, thousandths of a second). Under 10 ms, people barely notice any delay. Past 100 ms, mouse clicks and screen changes start to feel noticeably sluggish.
Why Does Latency Happen?
Light and radio waves are incredibly fast, but they still have a physical speed limit. When a signal goes from Seoul across the Pacific to a server in the US, even at the speed of light through fiber-optic cable, a round trip takes around 100 milliseconds at minimum. This physical distance is the first reason latency exists.
The many pieces of network equipment along the way also take time. Before a signal reaches its destination, it passes through dozens of routers (network devices that find the path) and switches. Each one checks the data and hands it to the next hop, adding processing delay.
Finally, there's congestion. Just as cars line up at a highway toll booth, data piling up in a device's temporary buffer makes waiting time shoot up. Latency comes from distance, equipment processing speed, and network congestion, all added together.
A Bit More Precisely
When people talk about latency in online games or internet connections, they usually mean round-trip time (RTT). It's the whole round-trip travel time: your signal reaches the server, and the server's reply comes back to your computer. The ping test used to check a network connection measures exactly this round-trip time.
In some fields, like sensors and machine control, latency can mean only the one-way time from sending a signal to its arrival. So depending on the situation, you need to tell whether a number is one-way or round-trip.
Latency also isn't fixed. It changes from moment to moment. How much latency wobbles unpredictably with network conditions is called jitter. If latency is a little high but steady, video plays smoothly. If jitter is bad, the picture stutters and freezes.
Common misconceptions
Upgrading to an expensive 1 Gbps internet plan will always get rid of game lag.
A 1 Gbps plan widens your bandwidth (the width of the data pipe). It doesn't shorten the physical speed or distance data has to travel. If your distance to the server and the route stay the same, latency barely drops.
Where you meet it
In an online fighting game, you press the attack button and your character moves half a beat late.
On a video call with someone overseas, you talk over each other and hear each other a beat late.
A self-driving car that spots an obstacle and sends a brake signal has to cut latency down to a few milliseconds to stay safe.
In one sentence
Latency is the response time between data leaving and arriving, and it's different from bandwidth, which is the width of the data pipe.