A single request can leave log lines across five different services. Those lines are mixed with thousands of other requests happening at the same time. Timestamps and service names help, but they are not enough to show which log lines belong to the same request.

A correlation ID solves this by giving each request a unique ID. That same ID is copied into every log, message, and downstream call created by the request.

Instead of searching scattered logs by hand, you search for one ID and follow the request across the system.

This chapter explains how to create, pass, check, and use correlation IDs across services, queues, background jobs, logs, and traces.

Consider a checkout request that crosses several services:

Each service logs something:

Those seven lines look related because we placed them together. In production, they are mixed with thousands of other log lines from the same second. Without a shared ID, you match timestamps, hostnames, and guesses about the call flow.

This works on low-traffic systems but fails under load.