Chapter 1 · Practice 3 of 4
Cover the four golden signals for every service
Why it matters
The SRE book's chapter on monitoring distributed systems says that if you can only measure four things about a user-facing system, measure these:
| Signal | What it is |
|---|---|
| Latency | Time to serve a request. Track successful and failed requests separately, because fast errors make averages look better |
| Traffic | Demand on the system, such as requests per second |
| Errors | Rate of failed requests, explicit (HTTP 500) and implicit (a 200 with the wrong content) |
| Saturation | How full the service is, for the resource that runs out first. Systems often degrade before 100% |
These four give every dashboard the same first row, so anyone on call can read any service's health in seconds.
To check, open the main dashboard for each service and confirm all four are on it, with errors and latency split by endpoint or operation.
Avoid
- Dashboards that start with CPU and memory instead of user-facing signals