AWS Knowledge Hub

English

Tiếng Việt

English

Tiếng Việt

Appearance

Amazon ECS ​

Amazon Elastic Container Service (ECS) is a fully managed container orchestration service that makes it easy to deploy, manage, and scale containerized applications.

Overview ​

ECS eliminates the need to install, operate, and scale your own container orchestration infrastructure. It integrates seamlessly with AWS services and provides a secure, scalable platform for running containers.

Key Features ​

  • Fully Managed: No infrastructure to manage, automatic scaling
  • Multiple Launch Types: EC2 and Fargate launch types
  • Task Definitions: Declarative configuration for containers
  • Service Management: Long-running tasks with automatic restarts
  • Load Balancing: Integration with ALB and NLB
  • Auto Scaling: Automatic scaling based on CloudWatch metrics
  • Security: IAM roles, security groups, and encryption support
  • Networking: VPC integration with public/private subnets

Core Concepts ​

Clusters ​

A logical grouping of tasks or services. Clusters can use EC2 instances or Fargate.

Task Definitions ​

Blueprints for your applications that describe:

  • Container images to use
  • CPU and memory requirements
  • Networking configuration
  • IAM roles
  • Environment variables
  • Logging configuration

Tasks ​

Instances of a task definition running in a cluster. Tasks can run standalone or as part of a service.

Services ​

Maintains a specified number of running tasks simultaneously. Services handle:

  • Task placement
  • Load balancing
  • Auto scaling
  • Health checks
  • Rolling updates

Launch Types ​

EC2 Launch Type ​

  • Run containers on EC2 instances you manage
  • Full control over instance types and configuration
  • Cost-effective for consistent workloads
  • Requires cluster management

Fargate Launch Type ​

  • Serverless compute for containers
  • No EC2 instances to manage
  • Pay only for resources used
  • Automatic scaling and patching

Use Cases ​

  • Microservices: Deploy and manage microservices architectures
  • Batch Processing: Run batch jobs and scheduled tasks
  • Web Applications: Host web applications with load balancing
  • CI/CD Pipelines: Container-based build and deployment
  • Hybrid Workloads: Mix EC2 and Fargate in the same cluster

Integration with AWS Services ​

  • Application Load Balancer: Distribute traffic across tasks
  • CloudWatch: Monitoring, logging, and metrics
  • IAM: Fine-grained access control
  • VPC: Network isolation and security
  • ECR: Container image registry
  • Auto Scaling: Automatic capacity management

Best Practices ​

  • Use Fargate for most workloads to reduce operational overhead
  • Implement health checks for reliable service management
  • Use task placement strategies for optimal distribution
  • Enable CloudWatch Container Insights for monitoring
  • Implement proper IAM roles with least privilege
  • Use secrets management (Secrets Manager, Parameter Store)
  • Configure log aggregation to CloudWatch Logs
  • Implement auto scaling based on CPU/memory metrics