CustomizedLoadMetricSpecification

Represents a CloudWatch metric of your choosing that can be used for predictive scaling.

For predictive scaling to work with a customized load metric specification, AWS Auto Scaling needs access to the Sum and Average statistics that CloudWatch computes from metric data.

When you choose a load metric, make sure that the required Sum and Average statistics for your metric are available in CloudWatch and that they provide relevant data for predictive scaling. The Sum statistic must represent the total load on the resource, and the Average statistic must represent the average load per capacity unit of the resource. For example, there is a metric that counts the number of requests processed by your Auto Scaling group. If the Sum statistic represents the total request count processed by the group, then the Average statistic for the specified metric must represent the average request count processed by each instance of the group.

If you publish your own metrics, you can aggregate the data points at a given interval and then publish the aggregated data points to CloudWatch. Before AWS Auto Scaling generates the forecast, it sums up all the metric data points that occurred within each hour to match the granularity period that is used in the forecast (60 minutes).

For information about terminology, available metrics, or how to publish new metrics, see Amazon CloudWatch concepts in the Amazon CloudWatch User Guide.

After creating your scaling plan, you can use the console to visualize forecasts for the specified metric. For more information, see View scaling information for a resource in the Scaling Plans User Guide.

Contents

See Also

For more information about using this API in one of the language-specific AWS SDKs, see the following: