Skip to main content

ECS Architecture

Container orchestration with Fargate and EC2 capacity providers, internal/external ALBs, and Service Discovery.

Problems this Architecture solves​

  • Standardizes how containerized services are exposed, discovered, and scaled in ECS.
  • Reduces operational overhead for teams that need managed runtime options without building cluster primitives from scratch.
  • Centralizes secrets, parameters, logs, and deployment behavior around a repeatable service pattern.

Key Features​

  • Fargate: Serverless compute, no EC2 management
  • Service Discovery: Cloud Map for DNS-based service discovery
  • External ALB: Public-facing HTTPS with WAF/Shield
  • Internal ALB: Private service-to-service communication
  • Auto Scaling: Target tracking based on CPU/memory/ALB metrics
  • Blue-Green Deployments: CodeDeploy integration for zero-downtime

Capacity Providers​

FARGATE​

  • Serverless compute, pay per vCPU/memory
  • No infrastructure management
  • Best for: Most workloads

FARGATE_SPOT​

  • Up to 70% cost savings
  • Can be interrupted with 2-minute warning
  • Best for: Fault-tolerant batch jobs

EC2​

  • Self-managed instances with ECS agent
  • More control over instance type/AMI
  • Best for: GPU workloads, reserved capacity

EC2_SPOT​

  • Up to 90% cost savings
  • Can be interrupted
  • Best for: Stateless, fault-tolerant workloads

Service Configuration​

api-service​

  • Tasks: 3 (high availability)
  • CPU: 512 (0.5 vCPU)
  • Memory: 1024 MB
  • Load Balancer: External ALB
  • Health Check: /health endpoint

worker-service​

  • Tasks: 2
  • CPU: 1024 (1 vCPU)
  • Memory: 2048 MB
  • Load Balancer: Internal ALB
  • Trigger: SQS queue

scheduler-service​

  • Tasks: 1
  • CPU: 256 (0.25 vCPU)
  • Memory: 512 MB
  • Schedule: EventBridge cron
  • Purpose: Background jobs

Supporting Services​

  • ECR: Private container registry with image scanning
  • Secrets Manager: Automatic rotation for DB credentials
  • Parameter Store: Environment variables and feature flags
  • CloudWatch: Logs, metrics, and alarms for monitoring