Pod Affinity in Kubernetes: What It Is and How It Works

In Kubernetes, pod affinity lets you tell the system to place pods close to other pods based on labels. It helps pods run on the same or nearby nodes to improve communication or resource sharing.

How It Works

Pod affinity is like asking Kubernetes to seat friends together at a dinner table. You define rules that say, "Put this pod near pods with certain labels." Kubernetes then tries to schedule pods on nodes where those other pods already run.

This helps pods that need to talk a lot or share resources stay close, reducing network delays and improving performance. If the preferred nodes are full or unavailable, Kubernetes may schedule pods elsewhere, but it tries to honor your affinity rules first.

Example

This example shows a pod spec with podAffinity that prefers to run on nodes where pods with label app: frontend exist.

yaml

apiVersion: v1
kind: Pod
metadata:
  name: example-pod
  labels:
    app: backend
spec:
  affinity:
    podAffinity:
      preferredDuringSchedulingIgnoredDuringExecution:
      - weight: 100
        podAffinityTerm:
          labelSelector:
            matchExpressions:
            - key: app
              operator: In
              values:
              - frontend
          topologyKey: kubernetes.io/hostname
  containers:
  - name: backend-container
    image: nginx

Output

Pod 'example-pod' will be scheduled preferably on nodes where pods labeled 'app=frontend' are running, improving proximity.

When to Use

Use pod affinity when your pods benefit from being close to certain other pods. For example:

• Microservices that communicate frequently and need low latency.
• Pods sharing local storage or cache.
• Workloads that perform better when grouped on the same node or rack.

This helps improve performance and resource efficiency in your cluster.

Key Points

• Pod affinity schedules pods near other pods based on labels.
• It uses preferredDuringSchedulingIgnoredDuringExecution or requiredDuringSchedulingIgnoredDuringExecution rules.
• Improves communication speed and resource sharing.
• Works by matching pod labels and node topology keys.
• Helps optimize cluster performance and pod placement.

Key Takeaways

Pod affinity helps schedule pods close to other pods with specific labels to improve communication.

It uses label selectors and topology keys to decide pod placement on nodes.

Use pod affinity for workloads that benefit from low latency or shared resources.

Affinity rules can be preferred or required, affecting scheduling strictness.

Pod affinity optimizes cluster resource use and application performance.