A company deploys an API by using an Application Load Balancer (ALB) that targets an...

The failures come from three separate pressure points: database connection exhaustion, CPU-bound Lambda execution, and high-volume S3 writes to one prefix. RDS Proxy is designed to improve database connection management, scalability, availability, and security, so it addresses the Aurora connection quota issue. Increasing Lambda memory is also correct because AWS allocates CPU proportionally with memory; CPU-bound functions can improve dramatically when memory is increased. For S3, AWS documents at least 3,500 PUT/COPY/POST/DELETE requests per second per partitioned prefix and notes that write performance can be scaled by writing across multiple prefixes. NLB in front of ALB adds no value, API Gateway does not solve the listed bottlenecks, and an Aurora read replica does not help a write-heavy workload.

================