Amazon EKS — Distributed Tracing

less than 1 minute read

Distributed Tracing

Tracing follows a request through multiple microservices, showing latency at each step. AWS X-Ray and OpenTelemetry are the main options on EKS.

Why Tracing Matters

Without Tracing:
User: "The app is slow"
You: "Which service? I have 20."

With Tracing:
User → API Gateway (5ms) → Auth Service (10ms) → User Service (500ms!)
                                                   ↑ Found it!

AWS X-Ray

# X-Ray DaemonSet
apiVersion: apps/v1
kind: DaemonSet
metadata:
  name: xray-daemon
spec:
  template:
    spec:
      containers:
        - name: xray-daemon
          image: amazon/aws-xray-daemon
          ports:
            - containerPort: 2000
              protocol: UDP

OpenTelemetry is the vendor-neutral standard. AWS Distro for OpenTelemetry (ADOT) is the recommended collector:

$ aws eks create-addon --cluster-name demo --addon-name adot \
    --service-account-role-arn arn:aws:iam::123456789012:role/ADOT-Role

Common Mistakes

Mistake Impact
Not instrumenting all services Gaps in trace data
Tracing 100% of requests High cost and storage
Not correlating traces with logs Can’t debug effectively

What’s Next?

Get notified: Alerting.

You may also enjoy