Serverless observability with OpenTelemetry
summary
Shows how to instrument Python Lambda functions with AWS Distro for OpenTelemetry using a prebuilt Lambda layer and the OpenTelemetry collector as a Lambda extension, exporting to AWS X-Ray. Demo scrapes a web page and saves to S3; initial uninstrumented run shows ~12s end-to-end (700ms init, ~11s invocation). Auto-instrumentation reveals a 10s sleep causing the delay; removing it drops execution to ~4s, with HTTP GET ~0.5s and S3 put ~277ms.
AI-written summary · about
chapters (18)
- 0:00Serverless observability intro
- 2:13Distributed systems challenges
- 3:39Observability pillars
- 5:26Context propagation in tracing
- 7:13Why OpenTelemetry
- 8:40Instrumentation methods
- 11:08OpenTelemetry architecture
- 13:56OpenTelemetry benefits
- 14:58AWS Distro for OTel
- 16:44Lambda layers and extensions
- 19:37OpenTelemetry in AWS Lambda
- 21:17Demo setup
- 23:09Troubleshooting without tracing
- 25:19Enabling X-Ray tracing
- 28:33Auto-instrumentation with OTel
- 30:45Analyzing traces
- 33:19Identifying bottlenecks in code
- 34:05Final results
related talks
01h 02mPython Powered Serverless Observability
43mObservability for Serverless
34mObservability in Serverless Application
30mUse the Observability, Luke!
24mHow To Monitor and Troubleshoot Applications using OpenTelemetry
41mLive Coding on Observability for Serverless
59mModernize faster with serverless and end-to-end observability
44mServerless – Automated Tests of web apps in the Cloud world