Malong RetailAI Protect can use Azure IoT Edge capability to deploy and manage the deep learning model for the solution, as shown in the following figure:
Figure 5. Managing the solution with Azure
Azure components include:
- IoT Hub—Used to provision, update, and manage IoT devices from the Azure cloud at scale. IoT Hub uses a security-enhanced channel to communicate from the cloud to the registered IoT devices.
- IoT Edge runtime—A collection of programs that turn a device into an IoT Edge device. Collectively, the IoT Edge runtime components enable IoT Edge devices to receive code or they enable a model to run at the edge and communicate the results. In this solution, the IoT Edge runtime is deployed in the PowerEdge server. IoT Edge runtime consists of two components (which are not shown in Figure 6).
- IoT Edge hub—Acts as a local proxy for IoT Hub by exposing the same protocol endpoints as IoT Hub. This consistency means that clients (whether devices or modules) can connect to the IoT Edge runtime similarly to connecting to IoT Hub.
- IoT Edge agent—Responsible for instantiating modules, ensuring that modules continue to run, and reporting the status of the modules back to IoT Hub. This configuration data is written as a property of the IoT Edge agent module twin.
- Container Registry—Used to store and distribute container images. When container images are registered, they can be deployed across the globe. Container Registry supports mechanisms for version control and for scanning and updating container images.
The Malong AI model is packaged as a container and is hosted on IoT Hub services on Azure using Container Registry services. IoT Edge Runtime is deployed on the PowerEdge server, which also includes Docker Runtime configured on the server. The server is then configured as an IoT Edge device and discovered by IoT Hub. When the PowerEdge server is configured as an IoT Edge device, the container with the AI model is pushed to the server from Azure IoT Hub. Model monitoring, updates, and management can now be performed from IoT Hub.