Observability in an Asynchronous World
summary
Claims that distributed tracing is the highest-value observability tool for event-driven systems and shows a working plant-based pizza demo that reproduces failures and fixes. Uses SQS and Lambda examples, adopts CloudEvents traceparent (W3C trace context) and recommends emitting event version and schema so downstream dependencies are discoverable. Shows span propagation vs span linking for long-running workflows, advises logging event schema (not payload) to avoid PII, and cites OpenTelemetry semantic conventions plus metrics like queue depth, message age and success rate.
AI-written summary · about
chapters (16)
- 0:00Intro
- 5:52What's an event?
- 6:50What does it mean to be "event-drive"?
- 7:49Events or messages
- 8:46Types of events
- 11:31Observability
- 13:48Challenges with observability in EDA
- 17:16Evolvability
- 21:09Now what?
- 26:23Documenting
- 31:07Tracing
- 37:46Metrics
- 40:27Success
- 40:55Observability in action
- 43:26Takeaways
- 46:02Outro
related talks
52mModern Observability and Event Driven Architectures
24mObservability across serverless asynchronous managed services
49mImproving Observability of Distributed Systems
01h 02mAdding Observability to Distributed Systems
18mObservability in Serverless Applications
01h 05mModern Observability and Event Driven Architectures
27mAchieving End-to-end Visibility into Complex Event-sourcing Transactions using Distributed Tracing
24mWhat is observability-driven development?