Question

A company operates a public REST API used by mobile clients worldwide. The API is served by an Application Load Balancer in front of Amazon ECS tasks in the us-east-1 Region. Approximately 80% of the requests are idempotent GET calls that return catalog data changing only a few times per day, while the remaining 20% are writes. International users report high and inconsistent latency, and the ECS fleet scales aggressively during peak hours to serve repetitive read traffic. The architecture team wants to reduce global read latency and offload repetitive reads from the origin while keeping the write path unchanged, with minimal application code changes. Which approach best meets these requirements?

Accepted Answer

Deploy Amazon CloudFront in front of the ALB, cache GET responses at edge locations using cache behaviors keyed on the catalog paths with an appropriate TTL, and forward non-cacheable methods to the origin.

Answer

Replicate the ECS service and ALB into multiple Regions and use Route 53 latency-based routing to direct all API traffic to the nearest Region.

Answer

Place an Amazon ElastiCache for Redis cluster in front of the ECS tasks and modify the application to check the cache before querying the catalog datastore.

Answer

Enable AWS Global Accelerator in front of the ALB to route users over the AWS backbone to the us-east-1 origin.