100% Free Forever
AI-Powered Learning
Industry Expert Content
Certificates & Badges
Learn At Your Own Pace
Cloud

AWS Fargate

By Amazon Web Services

IntermediateService12.8K learners

AWS Fargate is a serverless compute engine for containers that runs Amazon ECS or Amazon EKS workloads without requiring the user to provision, patch, or scale the underlying virtual machines.

Definition

AWS Fargate is a serverless compute engine for containers that runs Amazon ECS or Amazon EKS workloads without requiring the user to provision, patch, or scale the underlying virtual machines.

Overview

Before Fargate launched in 2017, running containers on AWS with Amazon ECS or Amazon EKS still meant managing a fleet of EC2 instances as the underlying capacity for those containers — patching operating systems, sizing instance types, and handling bin-packing of containers onto machines. Fargate removes that layer: you define a task or pod's CPU and memory requirements, and AWS runs the containers on infrastructure it manages entirely, billing based on the resources the task actually requests rather than the size of an underlying EC2 instance. Because Fargate is a launch type within ECS and EKS rather than a separate orchestrator, it fits into existing containerized architectures: a team can mix EC2-backed tasks for cost-sensitive, steady-state workloads with Fargate tasks for spiky or lower-operational-overhead workloads within the same cluster. This makes it a common middle ground between AWS Lambda, which is best for short event-driven functions, and self-managed EC2-based container clusters, which offer the most control but the most operational burden. Fargate is frequently chosen for workloads that need the flexibility of full Docker containers — longer execution times, custom base images, specific runtime dependencies — but where a team doesn't want to own the underlying compute fleet, which is a recurring theme across AWS's serverless service family.

Key Features

  • Serverless launch type for Amazon ECS and Amazon EKS
  • No EC2 instances to provision, patch, or scale manually
  • Billing based on requested vCPU and memory per task/pod
  • Supports standard Docker container images
  • Can run alongside EC2-backed tasks within the same cluster
  • Integrates with AWS networking, IAM, and logging services
  • Removes capacity planning and bin-packing concerns
  • Suited to longer-running workloads than typical FaaS functions

Use Cases

Running containerized microservices without managing servers
Batch processing jobs that need full container flexibility
Backend APIs packaged as Docker containers
Workloads with variable or unpredictable traffic patterns
Migrating containerized applications to AWS without managing EC2 fleets
Running Kubernetes pods on EKS without managing worker nodes

Frequently Asked Questions