Stop 12 · Monitoring & ObservabilityTopics · 3/8

Metrics, Logs & Traces

In short

When users say "the app is slow", you need three kinds of evidence: how often, what exactly happened, and where the time went. Those are metrics, logs and traces.

Metrics are numeric measurements over time, logs are discrete event records, and traces follow a single request across multiple services — together they give a complete picture of system behavior.

The blob character in a hospital bed beside a heart monitor as a nurse writes notes.

Think of it like this

The hospital ward

The bedside monitor shows heart rate as a live line. The nurse writes notes: "10:40, fever, gave paracetamol." The patient's file tracks their path from reception to X-ray to ward. Vitals, notes, route: metrics, logs, traces.

Watch it happen

Three signals, three questions

Observability

Metrics

p95 latency 2.1s

Logs

ERROR db timeout

Traces

payments 1.8s

1/5Three kinds of data, each cheap or expensive in different ways, each answering a different question.

The big ideas

What you need to know

Metrics

Counters, gauges and histograms with labels. Watch label cardinality: a user_id label creates millions of series.

Logs

Structured JSON logs with consistent fields are searchable. Free-text print statements mostly aren't.

Traces

A trace is a tree of spans, one per step, all sharing a trace ID passed between services.

OpenTelemetry

An open standard and SDKs for emitting all three, so you can switch backends without re-instrumenting.

See it in code

Walk through the code

1console.log(JSON.stringify({2  level: "error",3  msg: "payment failed",4  orderId: 42,5  traceId: "4bf92f3577b34da6a3ce929d0e0e4736",6  durationMs: 812,7  err: "db timeout after 800ms",8}));

1/3A level and a short, fixed message you can count and alert on.

One structured log line. Because traceId is a field, you can jump from this log to the full trace and back.

Save yourself the pain

Common mistakes

Oops

Logging every request in full detail at high traffic and calling it monitoring.

Do this instead

Use metrics for counts and rates, logs for details, and sample traces. Each costs differently.

Oops

Logging passwords, tokens or full card numbers while debugging.

Do this instead

Never log secrets or personal data. Mask fields before they ever reach the logger.

Your mission

Turn prints into signals

~30 min

Quick check

Your error-rate graph spikes. Which signal tells you which service in the request path was slow?

Go deeper

1Identify a real debugging scenario and which pillar (metrics, logs, or traces) would help most.
2Learn why traces are especially critical in microservice or serverless architectures.
3Understand these three pillars are complementary, not redundant with each other.