A web application uses CloudFront (S3 origin), API Gateway, Lambda, and an Aurora DB cluster. Under a large sales event, Lambda can handle peak requests but experiences latency on the initial burst because establishing DB connections is slow. Most Lambda invocations query the database. Which combination of actions will ensure scalability? (Choose three.)

To handle the initial burst of requests and reduce latency, increasing Lambda's provisioned concurrency is crucial. Provisioned concurrency keeps Lambda functions initialized and ready to respond immediately, avoiding cold starts which are a primary cause of latency during bursts. Converting the Aurora cluster to a global database with read replicas in regions near customers improves read scalability and reduces latency for geographically dispersed users by serving requests from closer replicas. Finally, using Amazon RDS Proxy for Aurora is essential for managing database connections efficiently. It pools and reuses connections, preventing the Lambda functions from overwhelming the database with new connection requests during high concurrency, which is a common cause of performance degradation. Increasing reserved concurrency only guarantees a minimum number of concurrent executions but doesn't prevent cold starts. Refactoring Lambda functions to initialize database connections inside the handler would worsen the problem by increasing the overhead for each invocation, rather than improving performance.