← Back to Docs
Recipe

APM Setup Primer

Instrument your stack with application performance monitoring so you catch regressions before users do.

Why APM Matters

APM gives you real-time visibility into latency, error rates, and throughput across every service. Without it, you are flying blind — debugging production issues becomes guesswork. A solid APM pipeline turns incidents into data-driven postmortems.

Pick Your Agent

Datadog

Broad language support, deep infrastructure correlation. Install the agent, enable APM, instrument with dd-trace.

OpenTelemetry

Vendor-neutral standard. Use the OTel Collector + your backend of choice (Honeycomb, Grafana, Jaeger).

Sentry

Error-first APM with tracing. Great for frontend + backend correlation on a single platform.

New Relic

All-in-one observability. Quick setup, strong dashboarding, generous free tier for small teams.

Instrumentation Checklist

  1. Deploy the agent — sidecar, daemonset, or library injection depending on your runtime.
  2. Enable distributed tracing — propagate trace context across HTTP, gRPC, and message queues.
  3. Tag everything — environment, service name, version, and region. Tags make dashboards useful.
  4. Set SLO-based alerts — alert on p95 latency and error budget burn, not raw thresholds.
  5. Validate in staging — run a load test and confirm traces appear before shipping to prod.

Next Steps

Once APM is live, pair it with structured logging and real-user monitoring for a complete observability picture. Check the logging pipeline recipe for the next layer.