OpenTelemetry rewards a little discipline. These are the practices that keep it maintainable as a team and a codebase grow.
The practices
- Send everything to a Collector, then fan out to backends
- Use tail-based sampling for traces to keep the interesting ones
- Propagate context across services with W3C tracecontext
What good looks like
A Collector pipeline: receive OTLP, batch, export.
receivers:
otlp: { protocols: { grpc: {}, http: {} } }
processors:
batch: {}
exporters:
prometheus: { endpoint: 0.0.0.0:8889 }
otlphttp: { endpoint: https://tempo:4318 }
service:
pipelines:
traces: { receivers: [otlp], processors: [batch], exporters: [otlphttp] }
metrics: { receivers: [otlp], processors: [batch], exporters: [prometheus] }Anti-patterns to retire
- Sampling misconfigured, so you keep everything and blow up cost
- Mixing SDK versions and losing context propagation
- Exporting directly from apps instead of through a Collector
Want senior help putting OpenTelemetry to work - or a team bootcamp on it? CodeWizard runs Israel's largest DevOps community and embeds named senior engineers with a transparent hours log. Book a free consult or explore the Academy.