Operations

Distributed tracing

A request's path across services, captured as a tree of spans.

In plain terms

OpenTelemetry standardised the wire format. Sampling is the open problem at scale — head-based vs tail-based.

Origin

Google Dapper paper (Sigelman et al., 2010) was the canonical reference. Twitter Zipkin (2012) and Jaeger (2017) followed. OpenTelemetry (CNCF, 2019) unified the wire format.

Where it shows up in production

OpenTelemetry Vendor-neutral instrumentation. Standard across Datadog, Honeycomb, New Relic, Jaeger, Tempo.
Jaeger CNCF graduated tracing backend. Originated at Uber.
Google Dapper Internal Google system that proved distributed tracing scales. The 2010 paper is still the best primer.

Sources & further reading

Paper Sigelman et al. — Dapper (Google, 2010)
Docs OpenTelemetry docs

Found this useful?