Is It Time To Version Observability? (Signs Point To Yes)
summary
Argues that we need Observability 2.0: a single source of truth of arbitrarily wide structured log events from which you can derive metrics, logs, and traces, not separate metrics/logs/traces silos. Shows concrete wins: one canonical log per request, add trace/span IDs, store raw events and aggregate at read time, and dramatic cost math (a naive metrics heatmap can cost ~$63,000/month). Watch if planning a migration away from low-cardinality metrics.
AI-written summary · about
chapters (13)
- 0:00Intro
- 0:47Every company is now a tech company
- 4:57Fast feedback loops
- 7:46Brief history of observability
- 9:54Observability 2.0
- 14:48Meh-trics
- 17:36Who uses it & how?
- 19:30How you interact with production
- 20:44How you debug
- 23:00The cost model
- 30:07What to do to move towards Observability 2.0
- 32:01What got us here won't get us there
- 35:05Outro
related talks
39mIs It Already Time To Version Observability? (Signs Point To Yes.)
33mKeynote: Observability is due for a version change; Are you ready for it?
30mObservability is More than Logs, Metrics & Traces
31mObservability for Developers: How to Get There
35mBeyond Logs, Metrics and Traces. The Data Analytics Paradigm Shift in Observability — Dotan Horovits
36mObservability? It's a Data Analytics Problem, You Fool!
39mThe 5W’s of Observability with Practical Examples
48mObservability or How to Know the State of Your System