Which solution meets these requirements?
Use AWS Fargate to host the web application with backend Amazon RDS Multi-AZ DB instances.
Use Amazon API Gateway with an edge-optimized API endpoint, AWS Lambda for compute, and Amazon DynamoDB as the data store.
Use an Amazon Route 53 routing policy with geolocation that points to an Amazon S3 bucket with static website hosting and Amazon DynamoDB as the data store.
Use an Amazon CloudFront distribution that points to an Elastic Load Balancer with an Amazon EC2 Auto Scaling group, along with Amazon RDS Multi-AZ DB instances.
Explanations:
While AWS Fargate can provide a managed environment for hosting applications and Amazon RDS Multi-AZ offers high availability, it may not fully satisfy the requirement for regional low latency during peak hours and millisecond latency for data retrieval.
This option effectively uses serverless architecture with Amazon API Gateway for API management, AWS Lambda for compute (which automatically scales), and Amazon DynamoDB, which provides single-digit millisecond response times. It also benefits from high availability and regional low latency.
This solution uses Amazon S3 for static website hosting, which is not suitable for dynamic web applications that require backend processing. It may also not meet the requirements for low latency during peak hours due to potential cold start issues and the nature of S3 as a data store, which does not provide the required millisecond latency for retrieval.
Although this option involves an Elastic Load Balancer and an EC2 Auto Scaling group for handling traffic and scaling, it does not fully address the requirement of avoiding infrastructure management. Additionally, while Amazon RDS Multi-AZ offers high availability, it may not meet the millisecond latency requirement for data access.
I design that the answer is:
Use Amazon API Gateway with an edge-optimized API endpoint, AWS Lambda for compute, and Amazon DynamoDB as the data store.