Lets you automatically change the number of instances providing a particular compute workload using defined metrics and thresholds that determine if the platform adds or removes instances