Latency vs Throughput

Definitions

Two distinct performance measures that are routinely conflated and frequently traded against each other. Latency is the time a single operation takes end to end; throughput is the number of operations completed per unit time. A system can have excellent throughput and terrible latency, and often does so by design: batching amortises fixed costs across many items, raising throughput while making each individual item wait for the batch to fill. The classic illustration is a truck full of hard drives, which has enormous throughput and appalling latency. Queueing theory ties the two together, showing latency rising sharply as utilisation approaches capacity, which is why systems tuned for maximum throughput often exhibit unacceptable tail latency. Deciding which to optimise is an early architectural commitment.
In plain English: Latency is how long one thing takes; throughput is how many things you can do per second — they're different measurements.
Example: "The database query takes 50ms (latency) but we can run 1,000 of them concurrently for 20,000 queries/second (throughput)."