What's the difference between latency and throughput, and how would each show up for users?

“Think of a highway: latency is how long one car takes to make the trip, and throughput is how many cars get through per hour. Users feel each one differently.”

Try answering out loud for a few minutes before you open this.

1. Latency: how long one request takes, such as 300 milliseconds to load search results
2. Throughput: how many requests the system handles per second
3. High latency shows up as slow screens and abandoned searches even on quiet days
4. Low throughput shows up as errors or queues during peaks, such as a flash sale
5. They can trade off: batching work raises throughput but can add delay
6. PM angle: set targets for both and test them at expected peak traffic