Node.js Observability with OpenTelemetry

Observability gives you visibility into what your system is doing in production. With OpenTelemetry (OTel), you can standardize traces, metrics, and logs across services.

Core Signals

• Traces: request flow across services
• Metrics: time-series measurements (latency, throughput, error rate)
• Logs: event details and context

Why OpenTelemetry?

• Vendor-neutral instrumentation
• Unified semantic conventions
• Works with most backends (Jaeger, Tempo, Datadog, New Relic, etc.)

Basic Node.js Setup

const sdk = new NodeSDK({
  traceExporter: new OTLPTraceExporter({
    url: process.env.OTEL_EXPORTER_OTLP_TRACES_ENDPOINT,
  }),
  instrumentations: [getNodeAutoInstrumentations()],
})
sdk.start()

Instrument HTTP + Database

Auto-instrumentation can capture:

• inbound HTTP requests
• outbound HTTP/fetch calls
• PostgreSQL/MySQL/Redis operations

This helps identify where latency is spent (network, DB, external API, etc.).

Add Manual Spans for Business Logic

const span = tracer.startSpan('checkout.calculateTotals')
try {
  const result = await calculateTotals(cart)
  span.setAttribute('cart.items', cart.items.length)
  return result
} finally {
  span.end()
}

Manual spans make traces meaningful beyond framework-level events.

Propagation Across Services

Ensure traceparent headers are forwarded so traces remain connected end-to-end. Without propagation, each service appears as isolated spans.

Production Tips

• Sample intelligently (e.g., 10% baseline, 100% on errors)
• Tag spans with tenant/user-safe identifiers
• Redact PII in attributes/logs
• Set SLO-based alerts from metrics

Golden Dashboard

Start with a dashboard including:

• P50/P95/P99 latency
• error rate by route
• top slow spans
• upstream dependency health

Final Takeaway

Observability is not just toolingit is feedback for architecture decisions. OpenTelemetry gives a practical baseline to understand performance, reliability, and failure behavior in Node.js systems.