A public retail web application uses an Application Load Balancer (ALB) in front of EC2 instances across multiple AZs and an Amazon RDS MySQL Multi-AZ backend. Target group health checks use HTTP and point at the product catalog page. Auto Scaling maintains the web fleet based on ALB health checks. Recently the app suffered an outage; Auto Scaling repeatedly replaced instances, while web server metrics looked normal. Investigation showed the database was heavily loaded and queries were very slow. Which two changes together will address the issue and improve monitoring of overall application availability and functionality for future growth? (Choose two.)

The problem indicates that the web servers were healthy, but the database was overloaded, causing slow queries and application outages. Deploying an Amazon ElastiCache cluster (either Redis or Memcached) between the web application and the RDS database will significantly reduce the load on the database by caching frequently accessed data. This directly addresses the root cause of the database overload. Changing the ALB target group health check to a simple static HTML page ensures that Auto Scaling only replaces instances that are truly unresponsive at the web server level, preventing unnecessary instance cycling when the issue is upstream (like the database). Adding an Amazon Route 53 health check that monitors the product page provides end-to-end application availability and functionality monitoring, as it validates the entire application stack, including the database. CloudWatch alarms on this Route 53 health check will notify administrators of actual application failures. Creating read replicas for RDS would offload read traffic, but the application would need to be updated to use the reader endpoint, and it doesn't address write-heavy loads. A TCP health check for the ALB is too basic and doesn't validate the web server's ability to serve content. An RDS CloudWatch alarm for recovery is reactive and doesn't prevent the overload or provide end-to-end monitoring.