A company runs an ecommerce static website hosted on Amazon S3 with CloudFront. API Gateway invokes Lambda functions for user requests and order processing. Lambda functions store data in an Amazon RDS for MySQL cluster that currently uses On-Demand instances. DB usage has been consistent for 12 months. Recently the site experienced SQL injection and web exploit attempts. Customers report order processing latency increases during traffic peaks, and Lambdas often have cold starts. As the company grows, it needs scalable, low-latency performance during traffic peaks, lower database costs, and protection from SQL injection and web exploits. Which solution meets these needs?

The correct option addresses all requirements. Provisioned concurrency for Lambda directly reduces cold starts, improving latency during peak periods. Purchasing RDS Reserved Instances is a cost-effective strategy for consistent database usage, lowering database costs compared to On-Demand. Integrating AWS WAF with CloudFront provides a web application firewall to protect against SQL injection and web exploits, as CloudFront is already in front of the static website. Incorrect options: Increasing Lambda timeouts doesn't resolve cold starts or improve performance; it just allows functions to run longer. AWS Shield Advanced is for DDoS protection, not specifically SQL injection or web exploits, which WAF handles. Increasing Lambda memory might improve performance but doesn't guarantee cold start reduction. Migrating to Redshift is for analytical workloads, not an OLTP database like MySQL, and wouldn't be a suitable replacement for RDS for an e-commerce site. Amazon Inspector is a security assessment service, not a real-time protection mechanism like WAF. Migrating to Aurora Serverless would provide scalability but might not be the most cost-effective solution for consistent usage compared to Reserved Instances, and the question asks for lower database costs. While Aurora Serverless offers good scaling, Reserved Instances offer predictable savings for stable workloads.