Keynote: Observability is due for a version change; Are you ready for it?
summary
Argues that the industry is moving from metric/three-pillar Observability 1.0 to a breakingly different Observability 2.0 where a single source of truth is wide structured events with IDs, letting you slice raw traces, derive metrics and SLOs, and close feedback loops. Shows a cost example (capturing request latency can blow to ~$63,000/mo with naive metrics), claims teams can delete 90–95% of paging alerts, and recommends adopting OpenTelemetry and canonical, arbitrarily wide structured-blob logs to speed debugging and ship with confidence.
AI-written summary · about
chapters (16)
- 0:00Introduction and keynote setup
- 1:04Why engineering efficiency matters
- 3:34Feedback loops and team performance
- 6:26Closing the loop with observability
- 8:12Defining observability 1.0 and 2.0
- 11:27Observability 1.0 limitations
- 14:17Data storage in observability tools
- 16:43From reliability to development lifecycle
- 18:30Workflow changes in observability 2.0
- 20:57Debugging in observability 1.0 vs 2.0
- 23:01Democratizing debugging with 2.0
- 24:04The hidden costs of metrics
- 27:13Why observability 2.0 is worth it
- 28:35Adopting structured logs today
- 31:05The future of software engineering
- 32:49Call to action and conclusion
related talks
39mIs It Already Time To Version Observability? (Signs Point To Yes.)
35mIs It Time To Version Observability? (Signs Point To Yes)
29mObservability; a gentle introduction
15mKeynote: ...What Does the Future Hold for Observability?
27mOpening the Door to Observability
39mThe 5W’s of Observability with Practical Examples
35mBeyond Logs, Metrics and Traces. The Data Analytics Paradigm Shift in Observability — Dotan Horovits
07mKeynote: OpenTelemetry And The Future of Open Source Observability