AWS Certified Solutions Architect – Associate (SAA-C03) — Question 291

A company sells datasets to customers who do research in artificial intelligence and machine learning (AI/ML). The datasets are large, formatted files that are stored in an Amazon S3 bucket in the us-east-1 Region. The company hosts a web application that the customers use to purchase access to a given dataset. The web application is deployed on multiple Amazon EC2 instances behind an Application Load Balancer. After a purchase is made, customers receive an S3 signed URL that allows access to the files.

The customers are distributed across North America and Europe. The company wants to reduce the cost that is associated with data transfers and wants to maintain or improve performance.

What should a solutions architect do to meet these requirements?

Answer options

• A. Configure S3 Transfer Acceleration on the existing S3 bucket. Direct customer requests to the S3 Transfer Acceleration endpoint. Continue to use S3 signed URLs for access control.
• B. Deploy an Amazon CloudFront distribution with the existing S3 bucket as the origin. Direct customer requests to the CloudFront URL. Switch to CloudFront signed URLs for access control.
• C. Set up a second S3 bucket in the eu-central-1 Region with S3 Cross-Region Replication between the buckets. Direct customer requests to the closest Region. Continue to use S3 signed URLs for access control.
• D. Modify the web application to enable streaming of the datasets to end users. Configure the web application to read the data from the existing S3 bucket. Implement access control directly in the application.

Correct answer: B

Explanation

Amazon CloudFront reduces data transfer out (DTO) costs from AWS to the internet compared to direct S3 transfers, while simultaneously enhancing download performance by caching content at global edge locations near the customers. Transitioning to CloudFront signed URLs ensures that access control is maintained securely. Other options, such as S3 Transfer Acceleration or Cross-Region Replication, would increase overall costs, while streaming through EC2 would incur higher compute and data transfer overhead.