Skip to content
Pablo Rodriguez

Global Infrastructure

AWS Global Infrastructure

Section titled “AWS Global Infrastructure”

Overview

Section titled “Overview”
Global Scale

The AWS Global Cloud Infrastructure is “a secure, extensive, and reliable cloud platform, offering more than 200 fully featured services from data centers globally.”

Infrastructure Scale

Section titled “Infrastructure Scale”
  • 102 Availability Zones in 32 geographic regions around the world
  • Enables highly available, scalable, and flexible architectures near customers
  • Lower latency and increased performance

Core Infrastructure Components

Section titled “Core Infrastructure Components”

AWS Regions

Physical geographical locations with two or more Availability Zones

Availability Zones

One or more data centers designed for fault isolation

AWS Local Zones

Extensions of Regions placing services closer to population centers

AWS Data Centers

Where data resides and processing occurs

Points of Presence (PoPs)

Edge locations for reduced latency to customers

AWS Regions

Section titled “AWS Regions”

Key Characteristics

Section titled “Key Characteristics”
  • Definition: A Region is a geographical area consisting of two or more Availability Zones
  • Connectivity: Connected to multiple ISPs and private global network backbone
  • Isolation: Regions are isolated from one another for fault tolerance and stability

Region Management

Section titled “Region Management”
  • Regions introduced before March 20, 2019 are enabled by default
  • Regions introduced after March 20, 2019 (Asia Pacific Hong Kong, Middle East Bahrain) are disabled by default
  • You must enable these newer Regions before use
  • Some Regions have restricted access (e.g., AWS GovCloud for US government agencies)

Data Replication

Section titled “Data Replication”
  • You control data replication across Regions
  • AWS provides information about country and state where each Region resides
  • Select Regions based on compliance and network latency requirements
  • AWS products and services availability varies by Region

Availability Zones

Section titled “Availability Zones”

Structure and Design

Section titled “Structure and Design”
  • Each Availability Zone comprises one or more data centers
  • Some Availability Zones have as many as six data centers
  • No data center can be part of two Availability Zones

Fault Isolation Features

Section titled “Fault Isolation Features”
  • Each Availability Zone designed as independent failure zone
  • Physically separated in typical metropolitan Region
  • Located in lower-risk floodplains
  • Discrete, uninterruptible power supply and on-site backup generation
  • Fed by different grids from independent utilities
  • Redundantly connected to multiple tier-1 transit providers

Usage Recommendations

Section titled “Usage Recommendations”
  • Availability Zone is most granular level of specification for services like Amazon EC2
  • You are responsible for selecting Availability Zones where systems reside
  • AWS recommends replicating across Availability Zones for resiliency
  • Systems can span multiple Availability Zones
  • Design systems to survive temporary or prolonged Availability Zone failure

AWS Local Zones

Section titled “AWS Local Zones”

Purpose and Benefits

Section titled “Purpose and Benefits”

Local Zones “make it possible for you to run latency-sensitive portions of applications closer to end users and resources in a specific geography.”

Key Features

Section titled “Key Features”
  • Extension of a Region - not standalone infrastructure
  • Place AWS compute, storage, database, and select services closer to population centers
  • Deliver single-digit millisecond latency for specific use cases
  • Managed and supported by AWS

Use Cases

Section titled “Use Cases”
  • Media and entertainment content creation
  • Real-time gaming
  • Reservoir simulations
  • Electronic design automation
  • Machine learning (ML)

Technical Implementation

Section titled “Technical Implementation”
  • High-bandwidth, secure connection between local workloads and Region workloads
  • Seamless connection to full range of in-Region services
  • Same APIs and toolsets as regular AWS services
  • Provides elasticity, scalability, and security benefits of cloud

AWS Data Centers

Section titled “AWS Data Centers”

Operational Characteristics

Section titled “Operational Characteristics”
  • Foundation for AWS infrastructure
  • Location where actual data resides and data processing occurs
  • Typically has tens of thousands of servers
  • All data centers are online and serving customers

Reliability and Failover

Section titled “Reliability and Failover”
  • State-of-the-art, highly available data centers
  • Automated processes move customer data traffic away from affected areas during failures
  • Core applications deployed in N+1 configuration
  • Sufficient capacity for traffic load balancing to remaining sites during data center failure

Network Equipment

Section titled “Network Equipment”
  • Custom network equipment sourced from multiple Original Device Manufacturers (ODMs)
  • Customized network protocol stack
  • ODMs design and manufacture products based on AWS specifications

AWS Points of Presence (PoPs)

Section titled “AWS Points of Presence (PoPs)”

Network Scale

Section titled “Network Scale”

CloudFront uses global network including more than 410 PoPs:

  • 400 edge locations
  • 13 regional mid-tier caches

Component Types

Section titled “Component Types”

Edge Locations

Section titled “Edge Locations”
  • AWS data centers and servers located close to customers
  • Designed to deliver services with lowest latency possible
  • Ensure popular content can be served quickly to customers
  • Support services like Amazon Route 53, AWS Global Accelerator, and CloudFront

Regional Edge Caches

Section titled “Regional Edge Caches”
  • AWS data centers between origin server and edge location
  • Have longer cache duration than edge locations
  • Bring more content closer to customers even if not popular enough to stay at edge location
  • Increase efficiency and are transparent to end user
  • Used by default with CloudFront

Global Distribution

Section titled “Global Distribution”

Edge locations located in:

  • North America, Europe, Asia, Australia
  • South America, Middle East, Africa, China

Content Delivery Process

Section titled “Content Delivery Process”
  1. Regional edge caches absorb content not frequently accessed enough for edge locations
  2. Provide alternative to fetching content from origin server
  3. Content flows from origin → regional edge cache → edge location → end user

The AWS Global Infrastructure provides a comprehensive foundation for building resilient, high-performance applications that can serve customers worldwide with optimal latency and reliability.