Managing Amazon Auto Scaling

  • Amazon
  • Managing Amazon Auto Scaling

Auto scaling is a feature that is unique to the cloud.  It allows for the system to automatically spawn new instances to meet increased demand then terminate excess instances once the demand subsides.  This facility ensures that your environment will always have sufficient capacity to meet the demand while conserving on the costs of operation.

There are four use cases for auto scaling:

  1. Maintain a fixed number of running EC2 instances.  Hands-off, self-healing of environment, replacing down or defective instances automatically.
  2. Manually scale EC2 instances. Hands-on, growing or shrinking environment on an ad hoc basis.
  3. Dynamically scale EC2 instances. Hands-off, growing or shrinking environment based on system-generated alerts.
  4. Scheduled scaling of EC2 instances. Hands-off, growing or shrinking environment on a predetermined, scheduled basis.

To create an auto scaled environment, you will need to create the following:

  • An image from the prototype instance that you wish to clone.  
  • A keypair to associate with the cloned instances.
  • A security group or groups to associate with the cloned instances.
  • An ELB (elastic load balancer).
  • A launch configuration which defines the parameters of the instances to be cloned.
  • An auto scaling group which defines the parameters of the scaling processes.
  • Auto scaling policies to scale up and scale down.
  • Metrics or events that cause scaling policies to be enforced.

If you have the first four elements, you can begin to create an auto scaling launch configuration and group.