Amazon EKS — EKS on Fargate

1 minute read

EKS on Fargate

AWS Fargate is a serverless compute engine for containers. With EKS on Fargate, each pod runs in its own isolated microVM — you never see or manage worker nodes.

How Fargate Works

EKS on Fargate
┌──────────────────────────────────────────────────┐
│  EKS Cluster                                     │
│                                                  │
│  EC2 Node (traditional):                         │
│  ┌────────────────────────┐                      │
│  │  Node                  │                      │
│  │  ┌──────┐ ┌──────┐    │  Shared kernel,      │
│  │  │ Pod1 │ │ Pod2 │    │  shared resources     │
│  │  └──────┘ └──────┘    │                      │
│  └────────────────────────┘                      │
│                                                  │
│  Fargate (serverless):                           │
│  ┌──────────┐ ┌──────────┐ ┌──────────┐        │
│  │ microVM  │ │ microVM  │ │ microVM  │        │
│  │ ┌──────┐ │ │ ┌──────┐ │ │ ┌──────┐ │        │
│  │ │ Pod1 │ │ │ │ Pod2 │ │ │ │ Pod3 │ │        │
│  │ └──────┘ │ │ └──────┘ │ │ └──────┘ │        │
│  └──────────┘ └──────────┘ └──────────┘        │
│  Each pod has its own kernel and resources       │
└──────────────────────────────────────────────────┘

Create a Fargate Cluster

$ eksctl create cluster \
    --name fargate-demo \
    --region us-east-1 \
    --fargate

Or with config file:

apiVersion: eksctl.io/v1alpha5
kind: ClusterConfig
metadata:
  name: fargate-demo
  region: us-east-1

fargateProfiles:
  - name: default
    selectors:
      - namespace: default
      - namespace: kube-system

Deploy a Pod on Fargate

# Deploy (pods matching Fargate profile run on Fargate)
$ kubectl create deployment web --image=nginx --replicas=2

# Check — nodes show as "fargate-ip-xxx"
$ kubectl get pods -o wide
NAME                   READY   STATUS    NODE
web-7854ff8877-abc12   1/1     Running   fargate-ip-192-168-100-5.ec2.internal
web-7854ff8877-def34   1/1     Running   fargate-ip-192-168-101-8.ec2.internal

Notice: each pod runs on its own Fargate “node” — there’s a 1:1 mapping.

When to Use Fargate

Use Case Why Fargate?
Batch jobs / cron Pay only when running
Dev/test environments No idle node costs
Bursty workloads Scales instantly
Security-sensitive Pod-level VM isolation
Small teams No node management

Quick Reference

# Create Fargate-only cluster
$ eksctl create cluster --name demo --fargate

# Check Fargate profiles
$ eksctl get fargateprofile --cluster demo

# Deploy to Fargate
$ kubectl create deployment web --image=nginx

What’s Next?

Learn how Fargate selects pods: Fargate Profiles.

You may also enjoy