OpenTelemetry Without the Overwhelm: A FOSS Observability Primer

OpenTelemetry is a CNCF-graduated standard for collecting traces, metrics, and logs — and despite wide adoption, most teams still struggle to actually ship it. The instrumentation docs are long, the Collector config is cryptic, and the backend decision feels permanent.

This talk cuts through that by anchoring everything to one model: SDK → Collector → Backend. We instrument a real service from scratch — starting with auto-instrumentation, layering in manual spans where auto-instrumentation falls short — then build a working Collector pipeline and wire it to open-source backends (Grafana, Jaeger, or Prometheus). No proprietary agents. No SaaS lock-in.

Attendees leave with a working mental model and a decision framework for picking the right FOSS observability stack based on their team's scale, existing infrastructure, and operational maturity

1. A working mental model for OTel's three-layer architecture (SDK → Collector → Backend) that maps to real decisions, not just diagrams

2. How to instrument a service with auto-instrumentation and where it breaks down — so you know exactly when to reach for manual spans

3. How the Collector pipeline actually works: receivers, processors, exporters, and why the config feels hard until you see the structure

4. A side-by-side look at Grafana, Jaeger, and Prometheus as OTel backends — what each is good at and where each falls short

5. A decision framework for picking a FOSS observability stack based on your team's scale, existing infrastructure, and how much you want to operate yourself