Amazon EC2 Auto-Scaling Concepts

External

• Auto-scaling User Guide http://docs.aws.amazon.com/autoscaling/latest/userguide/

Internal

• Amazon EC2 Concepts

Overview

Auto-scaling is an Amazon EC2 functionality that insures that an optimal number of EC2 instances are started and maintained running, to handle the load of a specific application. The EC2 instances whose number is controlled this way are part of an auto-scaling group.

Auto-scaling allows for scaling plans, which control the number of active instances as the demand on the application increases or decreases.

With auto-scaling, an application gets better fault tolerance: auto-scaling can detect instance failure, terminate it and launch a new instance to replace it. Auto-scaling uses availability zones: if one availability zone becomes unavailable, auto-scaling will launch instances in another one, to compensate.

The application also gets better availability: the right amount of capacity is brought on-line (or shut-down) via auto-scaling.

Auto-Scaling Group

Auto-scaling ensures that the number of EC2 instances in an auto-scaling group does not go below a minimum size and above a maximum size. The group has a desired capacity.

Typically, a web application tier (web tier, app tier, database tier, cache tier) is implemented as an auto-scaling group.

Auto-Scaling Groups and Availability Zones

An auto-scaling group can span multiple availability zones, within a region