Latency
The delay between asking for something and the moment the answer begins to arrive.
Glossary
Throughput
How much work or data a system can handle over a given stretch of time.
In one line
How much work or data a system can handle over a given stretch of time.
Also calleddata rate
Throughput measures volume over time. A system can have low latency yet modest throughput, or high throughput with noticeable latency. The two describe different qualities of the same flow, and confusing them is a common source of misunderstanding.
Good drivers balance the two thoughtfully, deciding when to favour a quick response and when to favour moving as much as possible at once. Grouping work together often raises throughput, while answering each request the instant it arrives often lowers latency. Rarely can both be pushed to their limit at the same time.
Throughput is what matters most for large tasks. When a great deal of data must move, the steady rate at which it flows decides how long the whole job takes, far more than how quickly the very first piece arrives.
Raising throughput often means working in batches and keeping every part busy without gaps. A pathway that is always carrying something, rather than starting and stopping, moves far more over the course of a task, which is why steady, uninterrupted flow is so prized.
In practice
In practice, throughput is like the number of cars a motorway carries each hour. A single car might travel quickly, but what decides how many people get through is the steady volume the road can sustain. Software measures its bulk work the same way: total amount delivered over time.
