Distributed Tracing
Tracing paths across async boundaries
Propagate context through queues, batch jobs, and hand-written HTTP clients without turning every span into noise.
- Format
- Cohort with async labs
- Duration
- 6 weeks, cohort
- Skill
- Advanced
- Stack
- OpenTelemetry, Jaeger
Tuition (informational): KRW 540,000
Request a syllabus conversation
Distributed tracing only helps when spans tell a story. This course spends time on async handoffs, fan-out fan-in patterns, and the humility of marking unknown gaps honestly. You instrument a reference app, break it on purpose, and compare trace readability before and after small naming and attribute tweaks.
What is included
- OpenTelemetry SDK setup across two services and a worker
- Exercises on span naming that survive refactors
- Baggage versus attributes decision trees
- Tail-based sampling discussion with trade-off math kept human-scale
- Debugging a dropped context through a message broker
- Annotating external vendor gaps without fictional precision
- Group critique of traces from anonymized incidents
Outcomes
- You can narrate a request path across three hops using one trace screenshot.
- You add instrumentation MRs that reviewers accept without nitpicking naming.
- You know when to stop adding spans and fix logging instead.
Instructor of record
Observability instructor with a background in JVM services and cranky message brokers.
Mira Cho
Primary feedback on labs
Participant questions
Reference services are Java and Go, but OTel concepts port widely. Polyglot questions are welcome.
Recent voices
“Fan-in lab made me rename spans that sounded poetic at 2pm and useless at 2am.”