Posted On: Mar 14, 2023

Application Auto Scaling customers can now use Amazon CloudWatch Metric Math to customize the metrics they use with Target Tracking policy without actually publishing and paying for the customized metrics separately. Customers can use both arithmetic operators (such as +, -, /, and *) and mathematical functions (such as Sum and Average) to easily create custom metrics based on existing CloudWatch metrics. Application Auto Scaling offers support to automatically scale capacity of 13 supported services, including Amazon Elastic Container Service (ECS) services. Specifically, Target Tracking works like a thermostat - it continuously changes the capacity of the scaled resource to maintain the specified metric at a customer-defined target level. Today’s release makes it easier and cheaper to configure Target Tracking with custom metrics.

Target Tracking offers out-of-the-box support for the most common metrics such as CPU Utilization of ECS services. In some cases, customers want to scale on their own application-specific metrics, such as the number of requests served, or on metrics published by other Amazon Web Services products, such as Amazon SQS. Until today, you would have to create custom CloudWatch metrics for Target Tracking to consume. Now, if the custom metric is a simple function of other existing metrics, you can use CloudWatch Metric Math in the Target Tracking policy, instead of publishing (and paying for) a new custom CloudWatch metric. For example, to define a custom metric representing the SQS messages per task in a ECS service, you could take the existing SQS metric for messages in queue and simply divide it by capacity in the Target Tracking Policy configuration using Metric Math to make it work with your Target Tracking policy.

Target Tracking is available through Amazon Command Line Interface (CLI) and Amazon SDKs in Amazon Web Services China (Beijing) Region, operated by Sinnet and Amazon Web Services China (Ningxia) Region, operated by NWCD. To learn more, visit the Application Auto Scaling documentation.