What is latency?

Latency is the waiting time between asking for something and the answer starting to arrive. It is not the same as connection speed. Speed is how much data arrives per second once things are flowing; latency is how long before the first piece shows up. The distinction matters because the fixes are different. More capacity moves more data at once, but only shorter distances or fewer stops make the answer come sooner.

Latency is the delay between a request and the start of the response, and it is a different measurement from bandwidth. Bandwidth counts how much data moves per second; latency counts the time before any of it arrives. A connection can have enormous capacity and still answer late, because carrying more at once is not the same as arriving sooner. Two things create the delay. The first is distance, because signals take real time to travel. A trip to a server across an ocean and back takes a noticeable fraction of a second, and no upgrade removes that, because nothing travels faster than light. The second is stops, because a request rarely goes anywhere directly. It passes through routers and servers, and each one spends a moment handling it before passing it on. The reason milliseconds matter is that they multiply. Loading one page involves dozens of little round trips, and each one pays the delay again. In a video call or a game you feel every delay the moment it happens. That is why the fixes for latency all take the same two shapes: move the two ends closer together, or make fewer trips between them.

Latency is how long a letter takes to arrive; bandwidth is how many letters fit in the van. The postal service can add vans forever, and a mountain of mail will move at once. Yet a letter to the next continent takes just as many days as before, because extra vans do not shorten the road. Only two things bring a reply sooner: moving closer, or settling the matter in two letters rather than six. A bigger van and a shorter road are different purchases.