Skip to content

Coming soon

Observability

How a production system becomes understandable from the signals it already emits.

A full guide isn’t written yet

The note below is the scope of this section. It isn’t a finished guide, and it doesn’t pretend to be.

Observability, as this section will use the word, is the ability to answer a new question about production from signals you already have. Monitoring, in the older sense, is checking the bounds you thought to define ahead of time. You want both.

The Site Reliability Engineering book treats four signals as the baseline for monitoring: latency, traffic, errors, and saturation. A guide here will eventually cover how those differ from logs and traces, and how to keep the cost of collecting them from becoming its own incident. That guide isn’t written yet.

Until it is, the public reference for those four signals is Monitoring Distributed Systems.