Beyond Logs, Metrics and Traces. The Data Analytics Paradigm Shift in Observability — Dotan Horovits
summary
Claims that observability is observability = data analytics not just logs+metrics+traces, demonstrates OpenTelemetry’s pipeline (receivers/processors/exporters) and cites concrete standards work: OpenMetrics, ECS merging into OpenTelemetry semantic conventions, and continuous profiling being added. Gives actionable practices: use structured logs with trace IDs and exemplars, roll up low‑value logs into metrics, limit retention, and reduce high‑cardinality labels. Shares hard numbers: 99% of index data never searched and 75% of stored metrics unused, arguing these drive cost and design choices.
AI-written summary · about
chapters (12)
- 0:00Introduction to observability
- 4:20Why three pillars fail
- 7:32Data analytics paradigm shift
- 8:15Beyond three pillars of telemetry
- 12:13OpenTelemetry overview
- 16:31Structured logs matter
- 19:42Open standards update
- 20:47Enrichment and correlation
- 23:15Unified query and visualization
- 25:04Data volume and noise ratio
- 30:07Key takeaways summary
- 31:34Q&A session
related talks
36mObservability? It's a Data Analytics Problem, You Fool!
01mObservability
32mRules of Observability: The Data Edition
33mWhat Developers Should Know About Observability
11mObservability 2.0 - Much more than just logs,metrics and traces
48mAPI Observability by Design: Building Systems That Tell Their Own Stories
26mYou Can’t Fix What You Can’t See: Tales from the Observability Frontier
30mObservability is More than Logs, Metrics & Traces