3.1.2.8. Deploying Workloads in Multiple Regions for Global Scalability

3.1.2.8. Deploying Workloads in Multiple Regions for Global Scalability

Multi-region deployment serves two purposes: disaster recovery and latency reduction. The architecture differs significantly for each use case.

Latency reduction architecture:

• CloudFront caches content at 400+ edge locations worldwide. For static content, this alone may be sufficient.
• Global Accelerator routes traffic over AWS's private backbone to the nearest healthy regional endpoint. Improves TCP/UDP latency by 60% vs. internet routing.
• Regional deployments with Route 53 latency-based routing direct users to the nearest application stack.

DR architecture (covered in detail in 3.1.3):

• Active-passive: Secondary region on standby, Route 53 failover routing
• Active-active: Both regions serve traffic, data replicated in real-time

Multi-region deployment automation:

• CloudFormation StackSets: Deploy identical infrastructure across regions
• CodePipeline multi-region: Pipeline stages can deploy to different regions sequentially (deploy to us-east-1, test, then deploy to eu-west-1)
• ECR cross-region replication: Automatically replicate container images to deployment regions

Exam Trap: Global Accelerator provides static anycast IP addresses — useful when clients can't resolve DNS (IoT devices, gaming clients). CloudFront uses DNS-based routing. If the question mentions "fixed IP addresses" or "TCP/UDP traffic," the answer is Global Accelerator, not CloudFront.