OpenTelemetry Auto Instrumentation for Effortless Observability
summary
Demonstrates deploying a Node.js app on Kubernetes using eBPF+OpenTelemetry auto-instrumentation and an OTLP collector to capture traces, metrics and logs with zero code changes. Shows an N+1 DB query found and fixed in hours, golden signals that identify the CPU-hogging microservice, and a dramatic reduction in mean time to repair after connecting data to OpenSearch dashboards. Notes limits: limited custom spans, kernel/version constraints, and need for domain knowledge.
AI-written summary · about
chapters (8)
- 0:00OpenTelemetry and eBPF intro
- 3:52Lagging launch case
- 7:25Auto instrumentation with eBPF
- 12:00Deep dive into OpenTelemetry
- 15:30Auto instrumentation limits
- 18:48Post-launch challenges
- 20:12Centralizing logs with OpenSearch
- 22:00Key takeaways
related talks
45mSimplifying Instrumentation in complex systems at scale with Open Telemetry
34mNetwork Observability with eBPF and OpenTelemetry
55mCloud-Native Observability with OpenTelemetry
23mModern Application Debugging: An Intro to OpenTelemetry
18mOpenTelemetry and Observability: What, Why, and Why Now?
45mFrom Noise To Signal: Building Smart Observability Pipelines With OpenTelemetry
32mAutomatic logs, traces and metrics in K8s using OpenTelemetry & SigNoz
31mFrom Kernel To Cloud: Building a Complete Open Source Observ...