AWS Global Infrastructure

AWS Global Infrastructure

├── Regions (geographic areas)
│   ├── Availability Zones (isolated data centers)

├── Edge Locations (CDN caching)

├── Local Zones (near cities)

├── Wavelength Zones (in telecom 5G networks)

└── Outposts (AWS on-premises)

Regions:

What it is: A Region is a geographic area where AWS has multiple Availability Zones (AZs). Each region is independent for security, compliance, and fault isolation.

Example:

  • us-east-1 → Northern Virginia, USA

  • eu-west-2 → London, UK

Key Facts:

  • Regions are physically separated (hundreds of miles apart in some cases).

  • Each region has its own pricing and services (some services aren’t available in all regions).

  • You choose a region based on:

    • Latency → closer to your users = faster.

    • Compliance → e.g., GDPR in Europe.

    • Cost → some regions are cheaper than others.

Pro Tip (Portfolio Worthy): Keep multi-region architecture diagrams in your portfolio showing disaster recovery with Route 53 failover.

Availability Zones (AZs)

What it is: An AZ is one or more data centers within a region, each with independent power, cooling, and networking.

Example:

  • us-east-1a, us-east-1b, us-east-1c

  • All in the same region but physically separate.

Why it matters: If one AZ goes down (power outage, fire, fiber cut), others in the region stay up. AWS recommends running workloads in at least 2 AZs for High Availability.

Real-world analogy: Think of a region as a city, and each AZ as a separate neighborhood with its own power grid.

Edge Locations

What it is: Small AWS data centers close to users for low-latency content delivery. Used by Amazon CloudFront (CDN) to cache data.

Example:

  • You’re in London → CloudFront delivers your web content from a London edge location instead of waiting for it from us-east-1.

Why it matters: Improves speed for users around the world.

Local Zones

What it is: AWS infrastructure close to large cities to run low-latency workloads. Connects back to the parent region for full AWS service access.

Example:

  • AWS Los Angeles Local Zone for video editing studios needing ultra-low latency.

Wavelength Zones

What it is: AWS infrastructure embedded in 5G telecom networks for ultra-low latency mobile apps.

Example: Gaming, AR/VR apps with milliseconds latency.

Outposts

What it is: AWS hardware racks installed in your own data center to run AWS services on-premises but connected to AWS Cloud.

Key Points :

  • Always multi-AZ for critical workloads.

  • Use multi-region for disaster recovery.

  • Leverage edge locations for global performance.

  • Place compute close to data to reduce latency and cost.

  • Use private connectivity (AWS Direct Connect) for secure, high-speed enterprise links.

  • Always multi-AZ for critical workloads.

  • Use multi-region for disaster recovery.

  • Leverage edge locations for global performance.

  • Place compute close to data to reduce latency and cost.

  • Use private connectivity (AWS Direct Connect) for secure, high-speed enterprise links.

AWS Global Accelerator

  • What it is: A networking service that improves global application availability and performance by routing traffic through AWS’s global backbone network.

  • How it works:

    • You get static Anycast IPs (public).

    • User traffic enters the nearest AWS edge location.

    • AWS then routes it through its private backbone to the nearest healthy AWS Region endpoint.

Is it part of AWS Regional Infrastructure?

Not exactly. Here’s why

  • Region vs Global Services

    • Most AWS services are regional (e.g., EC2, RDS, S3 buckets live in a region).

    • Some are global services (e.g., IAM, Route 53, CloudFront, Global Accelerator).

  • Global Accelerator belongs to the global networking layer, not a single region.

    • It spans multiple AWS regions.

    • Its static IPs are global and don’t change if you switch regions.

    • However, the endpoints it routes to (EC2, ALB, NLB, etc.) are regional.

    In short:

  • Global Accelerator = Global networking service.

  • It sits above AWS regional infrastructure and connects users to regional endpoints through AWS’s private network.

AWS Regional vs. Global Services

Regional Services

  • Definition: Operate within a specific AWS Region and are isolated from other regions.

  • Key Characteristics:

    • Data Residency: Data and resources are confined to a single region.

    • Availability Zones: Utilize multiple Availability Zones within the region for high availability.

    • Examples: Amazon EC2, Amazon RDS, Amazon S3 (buckets), Amazon VPC.

Global Services

  • Definition: Span across all AWS Regions and are not confined to a single region.

  • Key Characteristics:

    • Global Reach: Accessible from any region, often with a global endpoint.

    • Control Plane: Some have a centralized control plane in a specific region (e.g., IAM in us-east-1), but their data planes are global.

    • Examples: Amazon CloudFront, AWS IAM, Amazon Route 53.

AWS Regional Services

Definition:

  • Services that exist inside a single AWS Region.

  • Data, resources, and endpoints are region-specific.

  • You must choose a region when creating or using them.

  • Isolation: If one region has an outage, other regions are unaffected.

Key Points:

  • Supports high availability within a region (with Availability Zones).

  • Can be made public or private, depending on configuration.

Category
Service
Public / Private Notes

Compute

EC2

Private if no public IP; public if you assign one or behind ALB/NLB

Compute

ECS / EKS

Private if tasks/pods are in private subnets

Storage

S3 (regional bucket)

Can restrict to VPC endpoint (private) or public via internet

Storage

EBS

Private (inside VPC)

Databases

RDS / Aurora / DynamoDB

Can be public or private depending on subnet

Networking

VPC

Private by default

Networking

Elastic Load Balancer (ALB/NLB)

Public if internet-facing; private if internal

Security

KMS

Regional; can be accessed privately via VPC endpoint

Security

Secrets Manager

Regional; private via VPC endpoint

Analytics

EMR, Redshift

Regional; endpoints can be private/public

Application

CloudWatch, CloudTrail

Regional logs, metrics; API can be public

AWS Global Services

Definition:

  • Services that span all regions automatically.

  • No need to choose a single region.

  • Manage users, policies, or traffic globally.

  • Often internet-facing or globally accessible, but can interact with private regional resources.

Key Points:

  • Typically used for identity, DNS, traffic routing, content delivery, global acceleration.

  • Not tied to Availability Zones or regional resources, but they integrate with them.

Category
Service
Public / Private Notes

Identity

IAM

Global; manages users/roles across all regions

Identity

AWS Organizations

Global; account management across regions

Networking

Route 53

Public DNS service; global routing

Networking

CloudFront

Public CDN; caches content globally

Networking

Global Accelerator

Public static IPs; routes to regional endpoints

Security

AWS WAF (CloudFront or ALB)

Global rules for web apps

Security

AWS Shield Advanced

Global DDoS protection

Monitoring

CloudTrail (multi-region)

Logs can be centralized globally

Application

SNS (can be global topic)

Public APIs by default

Application

Cognito

User pools are global, app integration is region-specific

Hybrid / Special Cases

Some services can be both regional and global depending on usage:

Service
Notes

S3

Bucket is regional; objects can be accessed globally if public

Lambda

Functions live in a region; API Gateway endpoint can be global

API Gateway

Regional endpoints (default) or edge-optimized (global via CloudFront)

DynamoDB

Tables are regional; Global Tables replicate across regions

Why this Matters

  1. Architecture Decisions: Regional for isolated workloads; global for identity, DNS, CDN.

  2. Disaster Recovery: Regional = replicate manually; Global = automatically accessible.

  3. Security: Regional services can be private in VPC; global services often require IAM policies for access control.

  4. Latency & Performance: Use global services (CloudFront, Global Accelerator) to reduce latency worldwide.

Last updated