ECS

Amazon Elastic Container Service (ECS) is a service that allows you to easily launch and manege Docker containers running on AWS compute.

• ECS can manage thousands of containers

• Containers are appropriately registered with chosen load balancers and they come online and go offline.

• Contains can have individual roles attached, making security a breeze

• Extremely easy to set up and scale to handle any workload

Deployment Options

There are three Amazon ECS (Elastic Container Service) deployment options.

Amazon ECS on AWS Fargate

• Serverless compute engine where AWS manages the underlying infrastructure

• You don't need to provision, configure, or scale clusters of virtual machines

• Best for applications that don't require specific customization of the underlying host

• Pay-per-use model based on vCPU and memory consumed by containers

• Ideal for microservices and applications with variable workloads

• Lower operational overhead but typically higher cost per container

Amazon ECS on Amazon EC2

• You manage the EC2 instances that form your ECS cluster

• Provides more control over infrastructure and instance types

• Better for workloads requiring specific instance configurations or GPU support

• More cost-effective for stable, predictable workloads with high utilization

• Requires more operational management (patching, scaling, security)

• Supports placement strategies and constraints for container placement

Amazon ECS Anywhere

• Extends ECS to run containers on your own on-premises infrastructure

• Supports both virtual machines and bare metal servers

• Useful for hybrid cloud scenarios and edge computing

• Provides consistent container orchestration across cloud and on-premises

• You maintain full control and responsibility for infrastructure

• Requires more operational overhead compared to cloud options

• Good for workloads with data residency requirements or existing hardware investments

Key Decision Factors:

1. Control requirements: Fargate offers least control but lowest maintenance, EC2 provides balance, Anywhere gives most control

2. Cost optimization: EC2 typically most cost-effective for stable workloads, Fargate for variable workloads

3. Operational overhead: Fargate requires least management, followed by EC2, then Anywhere

4. Infrastructure requirements: Consider specific hardware needs, compliance requirements, or existing investments

5. Scaling needs: Fargate provides seamless scaling, EC2 requires more planning, Anywhere depends on your infrastructure

Task vs Service

• A Task is a one-time run of containers

• A Service maintains a specified number of Tasks running continuously

• Services can integrate with load balancers and handle task replacement if tasks fail

For example, if you have a task definition for a web application that includes both a web container and a Redis container, each "task" would be a running instance of both these containers, deployed together as a unit.