AutoScalr

AutoScalr is an AutoScaling service that uses machine learning to save operational costs on autoscaled applications by determining the optimal set of spot instances spread across multiple spot markets and On-Demand instances that also meets the availability requirements of a production workload. You simply set the target spare capacity desired, the instance types, and the maximum exposure to a single spot market, and AutoScalr will constantly analyze your application's current and historical workload, and the current and historical spot prices to determine the most cost effective set of instances to use at each moment in time. If a more cost effective set of instances is determined, AutoScalr automatically makes the changes required to optimize costs. If AutoScalr detects or predicts that a spot instance will soon be lost due to price, it automatically spins up a replacement instance in another spot market, and/or an On-Demand instance as a fallback if necessary. AutoScalr also scales up more effectively than the normal Auto Scaling Group by eliminating the need for a cool-down period. Since it constantly tracks the current instance set, including recently launched instances still spinning up, it can respond to traffic changes quicker and launch additional instances if necessary, making it easier to use and more effective than the normal Auto Scaling Group which will wait till the cool-down period ends. Simplify your operations and start saving money today.