A trace tells the story between your services.

A distributed trace represents the work behind an operation as related spans. Each span records an operation, timing, and contextual attributes. Propagated context connects work across service boundaries, while span links can express relationships that do not fit a simple parent-child tree. Together these relationships help explain where time went and where a request failed.

OpenTelemetry

Why it matters

Especially useful for understanding work that moves between an HTTP request, a queue, a worker, and a callback.

More context

Start with meaningful operation boundaries, then propagate context through the transports you use. Inspect whether your traces represent actual causal relationships. A job identifier remains useful business context; it serves a different purpose from a trace identifier.

Keep in mind

Instrumentation alone does not guarantee complete traces. Sampling, missing propagation, and export failures can leave gaps.

Original read