AWS Global Infrastructure

The AWS Global Infrastructure is the physical network of data centers, networking, and content delivery components that AWS operates worldwide. It is designed to provide secure, reliable, and scalable cloud services with low latency and high availability. The infrastructure is composed of three main components: Regions, Availability Zones, and Edge Locations. Additionally, services like AWS Local Zones, Wavelength Zones, and AWS Outposts extend the infrastructure to meet specific latency, regulatory, and hybrid requirements.

An AWS Region is a geographical area that contains two or more Availability Zones. Each Region is completely independent from other Regions, with its own set of services, pricing, and compliance certifications. As of 2025, AWS has 33 launched Regions globally, with more planned. For example, us-east-1 (Northern Virginia) is the oldest and most heavily used Region, while eu-west-1 (Ireland) is a popular European Region. When you launch resources like EC2 instances or RDS databases, you must select a Region. Data does not automatically replicate across Regions unless you explicitly configure cross-Region replication (e.g., S3 Cross-Region Replication or Aurora Global Database).

Why multiple Regions? - Latency: Deploy resources close to your users to reduce network round-trip time. - Compliance: Keep data within specific legal boundaries (e.g., GDPR requires data in EU Regions). - Disaster recovery: Run workloads in multiple Regions to withstand a full Region outage. - Cost: Pricing varies by Region due to local electricity, real estate, and tax differences.

Each Region contains multiple Availability Zones — typically three or more (e.g., us-east-1 has six AZs: us-east-1a, us-east-1b, etc.). An AZ is one or more discrete data centers with redundant power, networking, and connectivity. AZs are physically separated by several kilometers (but within 100 km of each other) to prevent correlated failures from events like tornados or earthquakes. They are connected via high-bandwidth, low-latency fiber optic links, enabling synchronous replication between AZs for services like Amazon RDS Multi-AZ or Amazon ElastiCache.

Key characteristics: - Each AZ has independent power, cooling, and physical security. - AZs are isolated from failures in other AZs. - The link between AZs is typically less than 2 milliseconds of latency. - You cannot choose a specific AZ by name (e.g., us-east-1a) across accounts — AWS maps AZ names to different physical data centers per account to balance load.

Edge Locations are sites that AWS uses to cache content for faster delivery to users. They are primarily used by Amazon CloudFront (CDN) and AWS Lambda@Edge. There are over 600 Edge Locations and 13 Regional Edge Caches worldwide. Unlike Regions, Edge Locations do not run compute or storage services directly (except Lambda@Edge for lightweight functions). They act as a cache: when a user requests content, CloudFront serves it from the nearest Edge Location if cached; otherwise, it fetches from the origin (e.g., an S3 bucket or HTTP server) and caches it for future requests.

Regional Edge Caches are a middle layer between Edge Locations and the origin, reducing load on the origin by caching less popular content at a regional level.

Local Zones are extensions of AWS Regions that place compute, storage, and database services closer to large population centers that are not near an existing Region. They are designed for workloads that require single-digit millisecond latency. For example, AWS Local Zones are available in cities like Los Angeles, Boston, and Miami. They are connected to the parent Region (e.g., us-east-1) for management and networking.

Wavelength Zones are designed for 5G applications, embedding AWS compute and storage at the edge of telecommunications providers' networks. This enables ultra-low latency for applications like autonomous vehicles, live video processing, and industrial IoT. They are available through partners like Verizon and Vodafone.

AWS Outposts bring native AWS services, infrastructure, and operating models to virtually any on-premises data center or co-location space. You can run EC2, EBS, ECS, EKS, and RDS locally on Outposts, connected to the nearest AWS Region for management. This is ideal for workloads that require low latency to on-premises systems, local data processing, or data residency requirements.

AWS owns a private global network backbone that interconnects all Regions, AZs, and Edge Locations. This backbone provides high throughput and low latency, avoiding the public internet for inter-Region traffic. Services like AWS Direct Connect allow customers to connect their on-premises data centers to this backbone via dedicated private lines.

In a traditional on-premises data center, you have a single location with limited redundancy (often just two power feeds and backup generators). If that location fails, everything goes down. With AWS, you can spread your application across multiple AZs in a Region, achieving 99.99% availability for a well-architected workload. You can also deploy across multiple Regions for disaster recovery, which is impractical for most on-premises setups due to cost.

A company like 'ShopGlobal' runs an e-commerce website serving customers worldwide. They deploy their application in three AWS Regions: us-east-1 (primary), eu-west-1 (Europe), and ap-southeast-1 (Singapore). Within each Region, they use two Availability Zones for high availability. They use Amazon Route 53 latency-based routing to direct users to the closest Region. For static assets (images, CSS, JavaScript), they use CloudFront with Edge Locations to cache content globally. The product catalog database is deployed as Amazon Aurora Global Database, with a primary in us-east-1 and read replicas in the other two Regions for low-latency reads. This architecture costs more due to cross-Region data transfer and multiple database instances, but it provides sub-100ms page load times worldwide and can survive a full Region outage. Misconfiguration example: If they forget to enable CloudFront caching for dynamic content, each request goes back to the origin, increasing latency and origin load.