Stop 11 · Monitoring & ObservabilityTopics · 6/9

Distributed Tracing (Jaeger, OpenTelemetry)

In short

A checkout request touches the gateway, cart, payment and inventory services. It took 3 seconds. Which one was slow? Distributed tracing draws that one request's path with a timer on every hop.

Tracking a single request as it flows through multiple microservices, capturing timing at each hop — essential for diagnosing latency issues in distributed systems where logs and metrics alone can't show the full picture.

The mascot follows a parcel along a conveyor belt with a handheld scanner past courier stations.

Think of it like this

Courier parcel tracking

Your parcel gets scanned at every hub: picked up, Delhi hub, Nagpur hub, out for delivery, each with a time. When it's late, the scans show exactly where it sat for two days. A trace is that tracking history, but for one web request across your services.

Watch it happen

One trace ID across three services

Gateway

Checkout

Payment

Jaeger

traceparent: 00-4bf9…

1/6The gateway starts a trace and passes its ID in the W3C traceparent header.

The big ideas

What you need to know

Trace and spans

A trace is one request. Each span is one unit of work inside it, with a start, an end and attributes.

Context propagation

The trace ID must travel with every call, usually in the traceparent header. One service that drops it breaks the chain.

OpenTelemetry is the standard

One set of open APIs and SDKs for traces, metrics and logs, which can send to Jaeger, Grafana Tempo or vendors.

Sampling

Busy systems keep only a share of traces. Make sure errors and slow requests are among the ones you keep.

See it in code

Walk through the code

1// tracing.js: load first with  node --require ./tracing.js server.js2const { NodeSDK } = require("@opentelemetry/sdk-node");3const { getNodeAutoInstrumentations } = require("@opentelemetry/auto-instrumentations-node");4const { OTLPTraceExporter } = require("@opentelemetry/exporter-trace-otlp-http");5 6const sdk = new NodeSDK({7  serviceName: "checkout",8  traceExporter: new OTLPTraceExporter({ url: "http://localhost:4318/v1/traces" }),9  instrumentations: [getNodeAutoInstrumentations()],10});11sdk.start();

1/4This file must load before your app code, so it can patch libraries like http and express.

Run Jaeger locally with the official jaegertracing image, publishing ports 16686 (UI) and 4318 (OTLP over HTTP), then open localhost:16686.

Save yourself the pain

Common mistakes

Oops

Instrumenting only one service and expecting a full picture.

Do this instead

Tracing pays off when every hop propagates context. Start with the gateway and the busiest services.

Oops

Keeping 100% of traces on a high-traffic system.

Do this instead

Sample. Head sampling keeps a fixed share; tail sampling in the collector can keep all errors and slow requests.

Your mission

See your first trace

~40 min

Quick check

Spans from service B show up in Jaeger, but as separate traces from service A's spans. What's the likely cause?

Go deeper

1Read a conceptual explanation of how a trace ID propagates across service calls.
2Explore Jaeger's UI (via a demo or local setup) to see a sample distributed trace.
3Learn why OpenTelemetry has become the standard, vendor-neutral way to instrument applications for tracing.