How Testability Drives Observability
summary
Argues that asserting on distributed traces makes telemetry testable and steers better instrumentation, replacing fragile end-to-end setups with lightweight trace assertions. Demonstrates open source trace-test (MIT) with web UI and CLI against a Pokemon API backed by Postgres, RabbitMQ and external PokeAPI, asserting HTTP status 200, span attribute pokemon.id = 25, and a span duration check trace span duration <1s which reads 596 ms in the demo, and shows tests run in CI.
AI-written summary · about
chapters (10)
- 0:00Start
- 0:27What is observability?
- 1:03What is distributed tracing?
- 3:41Typical instrumentation lifecycle
- 5:41Observability is hard
- 7:57What is Trace-based testing
- 8:13Trace-based testing: compared to an e2e test
- 13:08Advantages of trace-based testing
- 20:46Trace-based testing demo
- 35:08Tracetest community
related talks
36mHow Testability Drives Observability
39mHow Testability Drives Observability
16mBreaking the Chain of Blame: How to Get True Test Observability?
24mWhat is observability-driven development?
43mObservability-Driven Development with Kubernetes
16mDistributed Tracing is Awesome, Testing is Not... Until Now
30mTesting in A Distributed Systems World
21mThe Road to Observability Everywhere at Monday.Com