A company runs analytics software on Amazon EC2 instances that process jobs for data uploaded to Amazon S3. Users report some submitted data is not being processed. Amazon CloudWatch shows EC2 CPU utilization is consistently at or near 100%. The company wants to improve performance and scale based on user load. What should a solutions architect do?

The correct solution addresses both the immediate performance bottleneck and future scalability. Routing incoming requests to Amazon SQS decouples the processing from the ingestion, buffering requests when the system is busy. Configuring an EC2 Auto Scaling group based on queue size allows the system to automatically scale out (add more instances) when the queue backlog grows, ensuring all data is processed efficiently. Updating the software to read from the queue completes this robust, scalable architecture. Creating a copy of the instance behind an ALB (Application Load Balancer) would distribute the load but wouldn't solve the underlying CPU bottleneck if each instance is still at 100%. An S3 VPC endpoint is for private S3 access and doesn't directly address EC2 CPU utilization or scaling. Modifying the instance type (vertical scaling) might temporarily alleviate the CPU issue but doesn't provide automatic scaling based on user load, which is crucial for handling variable demand.