You Can’t Fix What You Can’t See: Tales from the Observability Frontier
summary
Advocates build observability around OpenTelemetry and pragmatic rules: use tail sampling in the collector to keep all traces with 4xx/5xx or long spans and a 5–10% baseline, convert spans to metrics with the span metrics processor, and prefer zero-code instrumentation where teams lack bandwidth. Log consistently in OTLP or JSON, include trace ID in logs, and never log secrets. For metrics use Prometheus semantics, watch naming and cardinality, and integrate Prometheus + OpenTelemetry with actionable alerts and SLO-driven dashboards.
AI-written summary · about
chapters (8)
- 0:00Welcome and introduction
- 2:33Observability as a team sport
- 4:20Tracing with OpenTelemetry
- 8:12Tail sampling for cost control
- 11:47Structured logging best practices
- 17:05Metrics and intentional instrumentation
- 23:03Intentional observability setup
- 25:10Closing and community invites
related talks
22mShift Into an Observability Mindset with OpenTelemetry
42mFirst Steps with Observability: OpenTelemetry and You
35mBeyond Logs, Metrics and Traces. The Data Analytics Paradigm Shift in Observability — Dotan Horovits
02h 41mThe Science of Signals: Mastering Telemetry for Observability
55mBeginner's guide to observability
36mObservability For You and Me with OpenTelemetry
32mAntipatterns in Observability: Lessons Learned and How OpenTelemetry Solves Them
36mObservability? It's a Data Analytics Problem, You Fool!