An ecommerce site uses CloudFront, an Apache web server tier on EC2 instances in an Auto Scaling group, and an Amazon Aurora MySQL DB cluster. During a promotion users experienced errors and timeouts when adding items to carts. Some web servers were terminated before their logs could be collected, and Aurora metrics were insufficient for analyzing query performance. Which combination of actions improves visibility into application performance during peak traffic? (Choose three.)

Publishing Aurora slow query and error logs to CloudWatch Logs provides critical insights into database performance bottlenecks, helping identify inefficient queries during peak loads. Instrumenting the application with AWS X-Ray SDK allows end-to-end tracing of requests across EC2 instances and Aurora, pinpointing where delays occur within the application and database. Installing the CloudWatch Logs agent on EC2 instances ensures that Apache access and error logs are centrally collected, even if instances are terminated, which is crucial for analyzing web server behavior and errors during traffic spikes. Streaming logs to Kinesis is an option, but CloudWatch Logs is more directly integrated for log analysis and monitoring. CloudTrail focuses on API activity, not application or database performance. X-Ray does not directly consume Aurora performance benchmarking streams.