Skip to content

Infrastructure

Partition

  • A completely independent aws infrastructure

  • aws

  • aws-ch
  • aws-us-gov
  • aws-eu (coming soon)

Regions

  • A cluster of data centers in a geographic area

  • Things to consider before choosing a region

  • Data compliance: data governance & legal requirements
  • Proximity: reduces latency
  • Available services: not all servies are available in all regions
  • Pricing: prices vary from region to region

  • Examples

  • us-east-1 (Northern Virginia)
  • eu-west-3 (Paris)
  • sa-east-1 (São Paulo)
  • ap-southeast-2 (Sydney)

Availability Zones

  • Each region has some availability zones
  • Usually 3, min 2, max 6
  • The AZ is a (one or more) discrete data center with redundant power, networking and connectivity
  • The AZ is designed to be independent (good for cases of failure)
  • Different AZ within a region are connect with high bandwidth and ultra-low latency network

  • Example (ap-southeast-2)

  • ap-southeast2a
  • ap-southeast2b
  • ap-southeast2c

Points of Presence (Edge Locations)

  • It's like a region, but limited.
  • It offers a way to serve content to the end user with the higher proximity possible (and lower latency)
  • It's used for caching

Service Types

  • Zonal services
  • Control plane: Regional
  • Data plane: Zonal
  • RDS, EC2, EBS
  • Regional services
  • Control plane: Regional
  • Data plane: Regional
  • S3, SQS, DybamoDB
  • Global services
  • Control plane: Single Region
  • Data plane: Globally Distributed
  • IAM (cp in us-east-1), CloudFront, Route 53, Global Accelerator