Posted On: Jul 7, 2022

EC2 Auto Scaling now publishes predictive scaling policy’s forecasts as a CloudWatch metric, enabling you to analyze, monitor, and set alarms on the accuracy of predictive scaling. Predictive Scaling is a scaling policy that proactively increases the capacity of your Auto Scaling group ahead of predicted demand, improving the availability of your application while reducing the need to stay overprovisioned that otherwise would have increased your EC2 bill. As predictive scaling only increases the capacity for your Auto Scaling groups, applying it to your current scaling configurations strictly enhances your application availability. However, an inaccurate prediction can potentially increase your cost. Now, you can use the extensive list of CloudWatch features to measure accuracy of predictions, view forecasts using the familiar CloudWatch graphs, and also set automatic alarms and notifications when predictions are above your desired levels.

Amazon EC2 Auto Scaling is a service that helps you meet application demand by automatically adding or removing EC2 instances to an Auto Scaling group according to the conditions you define. It already publishes a wide range of metrics to Amazon Cloudwatch, which is an AWS native service to monitor the health of your infrastructure and applications running on AWS. Now, predictive scaling forecasts are published as CloudWatch metrics for the past timestamps. You can leverage various CloudWatch features like Metric Math to create accuracy metrics such as Mean Absolute Percentage Error (MAPE) that are commonly used to measure timeseries forecasting accuracy, view multiple metrics on a single graph to understand when and by how much scaling policies are changing the capacity of your groups, or create dashboards and alarms for more automated monitoring experience.

Predictive Scaling is available through AWS Command Line Interface (CLI), EC2 Auto Scaling Management Console, AWS CloudFormation, and AWS SDKs. To learn more, visit the documentation.