A song of Decentralization and Observability: Dance with open telemetry
summary
Shows a concrete OpenTelemetry deployment on AWS: replaces per-pod ADOT sidecars with a centralized AWS Distro for OpenTelemetry collector service registered via Cloud Map service discovery, used by Fargate tasks. Demo uses Python and Java SDKs, ADOT jar, DynamoDB and S3 calls, and AWS X‑Ray traces that surface a deliberately deleted schema; traces reveal parent/child spans and error codes so debugging happens in the console. Sampling 100% telemetry costs roughly $250 for X‑Ray plus ~$80 CloudWatch, total ~$300–320/month.
AI-written summary · about
chapters (10)
- 0:00Introduction and agenda
- 2:33What is observability
- 7:07Why open telemetry matters
- 11:52Open telemetry components
- 13:16Tracing in Python demo
- 15:43Production deployment patterns
- 18:11AWS open telemetry service demo
- 20:22Tracing with AWS X-Ray
- 24:34Cost and data considerations
- 26:03Q&A session
related talks
27mEnd-to-end visibility in the cloud w/OpenTelemetry
22mOpenTelemetry – An Observability Framework for Cloud-Native Software
22mOpen Telemetry for API Monitoring
24mObservability with AWS
35mOpenTelemetry for Busy People
57mYour Distributed System is Talking to You: OpenTelemetry
21mOpen Telemetry
01h 28mUnderstanding Distributed Systems with OpenTelemetry