Part 1 of 6 · 1 min read
Why this page comes first
In this module you will
- Describe a system's current load using concrete load parameters.
- Explain the difference between throughput and response time, and between latency and response time.
- Read a response-time distribution and explain why percentiles matter more than the average.
- Explain why high percentiles matter even when they affect a small fraction of requests.
- Reason about a fan-out problem the way a real large-scale system does.
Every earlier page in this subject assumed a database running comfortably under normal load. This page gives the vocabulary for talking precisely about load and performance. Every later page in this level, and every real production incident, actually uses it.
Remember this
The average response time hides the story; the high percentiles, like p99, are what a system's worst-treated users actually feel.