ECS clusters that have EC2 instances provisioned expose the CPUUtilization and MemoryUtilization CloudWatch metrics. Blue Matador monitors both of these metrics and will warn you if you are approaching 100% utilization in either resource. A cluster can function while near 100% utilization, but you will be unable to schedule more tasks on the EC2 instances in the cluster if there are not enough resources on any instance for a task to be scheduled.
Resource utilization can be lowered by either scaling the cluster by adding more EC2 instances, or moving some of the existing tasks to Fargate. Tasks can be ran in Fargate regardless of the ECS cluster setup by creating a new task definition with the Fargate launch type.