Amazon announced that SageMaker Edge Manager now integrates with AWS IoT Greengrass Version 2, allowing users to deploy ML model packages and Edge Manager agent (the on-device inference engine of Edge Manager) to fleets of edge devices easily.
Prior to this launch, the only way to deploy the Edge Manager agent to devices was to manually copy the Edge Manager agent to set up the devices and fleets for Edge Manager. For those who already had an AWS IoT Greengrass device, deploying a model package to the device using AWS IoT Greengrass would have required users to custom build a component from scratch and manually associate ther AWS IoT IAM role with the role alias generated by SageMaker Edge Manager.
With the integration of SageMaker Edge Manager and AWS IoT Greengrass V2, users can now get the Edge Manager agent as an AWS Greengrass V2 component. AWS IoT Greengrass components are software modules, such as app building blocks, libraries, or any code, that will run on an AWS IoT Greengrass device. Users can use a pre-built public Edge Manager component, which is maintained by AWS, or use a private Edge Manager component, which can be auto generated when users package their machine learning model with the CreateEdgePackagingJob API. Users can also create components for an edge application and inference logic, and manage the deployment of these components from AWS IoT console to devices. SageMaker Edge Manager also allows users to reuse existing AWS IoT role alias. If users do not have one yet, Edge Manager will generate a role alias as part of the Edge Manager packaging job. Users no longer need to associate a role alias generated from the SageMaker Edge Manager packaging job with AWS IoT Role.
Amazon SageMaker Edge Manager can be used with the integrated deployment capability in the following AWS regions: US East (N. Virginia), US East (Ohio), US West (Oregon), EU (Ireland), EU (Frankfurt), and Asia Pacific (Tokyo).