kubernetesnetworkingPractitioner

Mastering EKS Container Network Observability for Inter-AZ Traffic

5 min read AWS Containers BlogMay 5, 2026Reviewed for accuracy

Practitioner — Hands-on experience recommended

In a cloud-native world, visibility into network traffic is paramount. EKS Container Network Observability provides the tools you need to track inter-AZ and NAT gateway traffic, ensuring your applications run smoothly across availability zones. This capability is vital for diagnosing issues, optimizing performance, and maintaining high availability in your Kubernetes clusters.

At the core of this observability is the Network Flow Monitor Agent, an eBPF daemon that collects and monitors network traffic at the pod level. This allows you to see how traffic flows between your services, including insights on latency and throughput. You can also control traffic distribution using the trafficDistribution field in your Kubernetes Service spec, which guides kube-proxy on how to route traffic to service endpoints. For example, setting trafficDistribution: PreferSameZone helps keep traffic within the same availability zone, reducing latency and improving performance.

In production, you need to be aware of the Kubernetes version requirements—this feature is available from version 1.35 onward. Additionally, while using traffic distribution is beneficial, moving workloads to a single AZ is generally not recommended for critical applications that require fault tolerance. Always consider the implications of your architecture on availability and performance.

Key takeaways

→Leverage the Network Flow Monitor Agent to gain pod-level insights into network traffic.
→Use the `trafficDistribution` field to control how traffic is routed to service endpoints.
→Ensure your EKS cluster runs Kubernetes version 1.35 or later to utilize these features.
→Avoid moving workloads to a single AZ to maintain high availability and fault tolerance.

Why it matters

In production, understanding and optimizing network traffic can significantly enhance application performance and reliability, directly impacting user experience and operational efficiency.

Code examples

YAML

1apiVersion: v1
2kind: Service
3metadata:
4  name: receiver
5spec:
6  trafficDistribution: PreferSameZone
7  selector:
8    app: receiver
9  ports:
10    - port: 80

Bash

kubectl get pods -l 'app in (sender,receiver)' -o custom-columns='NAME:.metadata.name,APP:.metadata.labels.app,NODE:.spec.nodeName,AZ:.metadata.labels.topology\.kubernetes\.io/zone' --sort-by='.metadata.labels.app'

YAML

1apiVersion: v1
2kind: Service
3metadata:
4  name: my-service
5spec:
6  selector:
7    app: my-app
8  trafficDistribution: PreferSameZone
9  ports:
10    - protocol: TCP
11      port: 80
12      targetPort: 8080

When NOT to use this

Move your workload to a single AZ: This is generally not recommended for critical workloads that require high availability and fault tolerance.

Want the complete reference?

Read official docs

Test what you just learned

Quiz questions written from this article

Take the quiz →

Linux FoundationSponsor

Industry-standard certifications built by the people behind Linux and Kubernetes. Earn the CKA — the gold standard Kubernetes administrator cert. OpsCanary readers get 30% off year-round with code OPSCANARY3.

Get CKA certified →

Mastering EKS Container Network Observability for Inter-AZ Traffic

Key takeaways

Why it matters

Code examples

When NOT to use this

More on this topic

Deploying Dragonfly Lightweight: P2P Distribution Without the Database Overhead

K8gb: Elevating Global Load Balancing in Kubernetes

TCPRoute and UDPRoute in Gateway API v1.6: What You Need to Know