Observability: Superpowers for Developers
summary
Argues that folding observability into development makes production part of the feedback loop, using Honeycomb-style high-cardinality traces and logs so developers can ask real questions about real requests. Describes distributed tracing as “Chrome devtools for server code,” shows using feature flags with 3% traffic to test performance variants in production (Geckoboard ships multiple implementations and picks a winner in a day), and recommends instrumenting high-cardinality attributes like build_id and customer_id to speed debugging and reduce pages.
AI-written summary · about
chapters (14)
- 0:00The evolving role of developers
- 3:24Bridging dev and ops
- 5:09Observability as a new test
- 7:23The virtuous cycle
- 9:51Activating developer powers
- 12:52Instrumentation best practices
- 14:38Tracing and logs
- 17:03Debugging in production
- 18:55Data-driven development
- 21:47Testing in production
- 25:02Feature flags and analysis
- 26:49Social debugging
- 30:05Recap and cultural shift
- 33:04Q&A session
related talks
31mObservability for Developers: How to Get There
29mObservability for Developers: How to Get From Here to There
37mObservability-Driven Development
45mObservability is for User Happiness
23mA peek into observability from a tester's lens
53mObservability Driven Development
33mWhat Developers Should Know About Observability
50mObservability and the Development Process