Which scaling policy adjusts instances based on defined adjustments across various metric ranges?

Step Scaling is a scaling policy that allows you to set scaling adjustments that depend on the size of the change in the metric. This means that you can define specific thresholds that dictate how many instances to add or remove based on the metric's value. For example, if a metric like CPU utilization rises above a certain threshold, you can configure step scaling to add one instance at a minor increase, two instances at a higher increase, and so on. This granularity in scaling behavior allows for more responsive and efficient adjustment of resources based on real-time demands.

In contrast, simple scaling adjusts the number of instances based on a single threshold and typically responds with a fixed number of instances, without considering varying levels of metric values. Dynamic scaling is a broader term that can encompass various strategies but lacks the specific, tiered adjustments that step scaling provides. Target tracking scaling simplifies scaling decisions by automatically adjusting instances to maintain a desired metric value without the complexity of defining thresholds and adjustments. Thus, step scaling's ability to respond with specific adjustments to varying levels of demand makes it the correct choice for this scenario.