Observability for Emerging Infra: What Got You Here Won't Get You There
summary
Modern infra breaks dashboards and metrics; observability is the ability to ask new questions of raw, wide events not metrics and get answers. Build structured logs, traces and high-cardinality event storage (hundreds of dimensions) so any request can be inspected without shipping code. Do safe canaries and test in production; one real outage example is Android push failures caused by round‑robin DNS plus a router UDP/TCP edge case. With proper instrumentation teams high cardinality can delete 95% of alerts and keep only errors/latency/request‑rate.
AI-written summary · about
chapters (20)
- 0:00Introduction and context
- 1:11Defining software ownership
- 4:52Speaker background and bias
- 6:22Observability vs monitoring
- 14:55Managing system complexity
- 16:52The necessity of new tools
- 17:32System-level observability
- 20:07Debugging in modern environments
- 21:48Staging versus production
- 22:44Testing in production
- 25:53Observability as a missing link
- 26:10Real-world debugging examples
- 34:32Instrumentation conventions
- 37:16Cardinality and raw data
- 39:35The problem with aggregates
- 40:52The role of dashboards
- 41:19Formulating the right question
- 42:24The power of exploration
- 44:41Conclusion and rewards
- 46:15Final remarks
related talks
43mObservability for Emerging Infra: What Got You Here Won't Get You There
39mObservability for Emerging Infra: What Got You Here Won't Get You There
44mObservability for Emerging Infra: What Got You Here Won’t Get You There
35mObservability: Building Confidence & Answering the What, Where, How
06mIgnite: Observability Observed
29mObservability for Developers: How to Get From Here to There
52mObservability: it's not just an ops thing
29mObservability; a gentle introduction