Amazon SageMaker Monitoring
Last updated on 02 July, 2025LogicMonitor provides support for monitoring Amazon SageMaker. It helps you optimize your SageMaker service by enhancing visibility of CPU and GPU utilization, invocations, latency, and more.
Requirements for Monitoring Amazon SageMaker
To monitor Amazon SageMaker, you must configure monitoring of AWS DataSources, and ensure your AWS IAM policies and roles are updated to include SageMaker. For more information, see AWS Monitoring Setup.
Import Amazon SageMaker LogicModules
Install all Amazon SageMaker LogicModules from the LogicMonitor Module Exchange. If all SageMaker modules are already present, ensure the modules are using the most recent version.
For more information, see Module Installation.
AWS SageMaker LogicModules in Package
LogicMonitor’s package for AWS SageMaker consists of the following LogicModules. For full coverage, import the following LogicModules into your LogicMonitor platform:
Name | Type | Description |
AWS_SageMaker_Endpoint_Variant | DataSource | Monitors Amazon SageMaker endpoint performance metrics. |
AWS_SageMaker_Endpoint | DataSource | Monitors Amazon SageMaker endpoint performance metrics. NOTE: This DataSource is valid only for one variant SageMaker Endpoint named allTraffic and won’t work properly for multi variant. There is a new Datasource AWS_SageMaker_Endpoint_Variant which handles correctly all variant cases and it’s recommended to be eventually used for SageMaker |