A company operates a three-tier architecture for their online order processing system. The architecture includes EC2 instances in the web tier behind an Application Load Balancer, EC2 instances in the processing tier, and Amazon DynamoDB for storage. To decouple the web and processing tiers, the company uses Amazon Simple Queue Service (Amazon SQS). During peak demand, some customers experience delays or failures in order processing. At these times, the EC2 instances in the processing tier reach 100% CPU utilization, and the SQS queue length increases significantly. These peak periods are unpredictable. What should the company do to improve the application's performance?