Auto Scaling enables you to scale up or down the number of instances you are using based on parameters that you specify, such as traffic or CPU load.
Auto Scaling also monitors the health of each Amazon EC2 instance that it launches. If any instance terminates unexpectedly, Auto Scaling detects the termination and launches a replacement instance.
For a high degree of flexibility, you can organize Amazon EC2 instances into AutoScalingGroups, which enable you to scale different server classes (e.g., web servers, back end servers) at different rates. For each group, you specify the minimum number of instances, the maximum number of instances, and the parameters to increase and decrease the number of running instances.
For more information, refer to the Amazon Auto Scaling Developer Guide.