A Developer manages a monitoring service for a fleet of IoT sensors in a major city. The monitoring application uses an Amazon Kinesis Data Stream with a group of EC2 instances processing the data. Amazon CloudWatch custom metrics show that the instances a reaching maximum processing capacity and there are insufficient shards in the Data Stream to handle the rate of data flow. What course of action should the Developer take to resolve the performance issues?