The target being referred to is a target value for a specific metric that CloudWatch metrics monitor. The two ECS metrics CloudWatch monitors are CPU utilization and memory utilization. With target tracking scaling policies, the number of tasks is increased or decreased with the objective that the target value of a metric is kept. It may not be feasible to keep the precise target value of a metric because the number of tasks are increased/decreased in increments/decrements of 1. The metric is kept at the precise configured target value or close to the target value. A target tracking scaling policy itself adjusts itself to the fluctuations in a metric value to avoid rapid fluctuations in the number of tasks. For high application availability, a service scales out faster than it scales in, as fewer tasks hamper application availability while extra tasks do not. Multiple target tracking scaling policies may be set concurrently and the service is scaled to fulfill the scaling policy that effects the greatest increase in task capacity. Increases in CPU or memory utilization are indicators that the application is in need of more tasks and the service scales out if a target metric is exceeded. Sufficient metric data is a prerequisite for a scaling policy to scale. If insufficient data is available, the scaling policy does not scale the number of tasks. To scale if insufficient data is available, a step scaling policy should be used, which is discussed next.