Unified End-to-End Observability: How Comcast Generates SpanMetrics at Enterp...
summary
Comcast describes migrating 350 applications from AppDynamics to an OpenTelemetry stack that generates RED metrics by converting spans with the OpenTelemetry Collector and the SpanMetrics connector. The architecture uses an internal Conduit pattern with deterministic routing so all spans for one service go to a single writer, preserving global P99 correctness. A fixed‑seed smoke test verifies client and server counts match, and reusable Grafana dashboards plus label governance control cardinality and privacy.
AI-written summary · about
chapters (13)
- 0:00Trusted observability at scale
- 1:44Migrating to span metrics
- 3:10Span metrics for red signals
- 4:58P99 correctness with routing
- 6:00Conduit architecture
- 7:24Avoiding label explosion
- 8:06Seven migration gates
- 9:09Privacy and cost governance
- 10:33Demo setup validation
- 14:36Operator journey detect localize
- 21:24ClickHouse backend demo
- 24:16Reusable adoption blueprint
- 25:19Takeaways ownership labels hygiene
related talks
23mEnhancing Asynchronous Communication Observability with OpenTelemetry
05mFrom Observability to APM - Taking Cloud Native Up The Stack
27mEnd-to-end visibility in the cloud w/OpenTelemetry
15mExtending OpenTelemetry to Observability
26mScaling OpenTelemetry for Modern Workloads: From Mobile To Ma...
42mFirst Steps with Observability: OpenTelemetry and You
22mOpen Telemetry for API Monitoring
36mPanel: OpenTelemetry...