During the recent Connect(); conference, Microsoft announced a new version of the IoT Gateway Software Development Kit (SDK) called IoT Edge to provide edge computing in IoT scenarios.
IoT Edge uses modules to provide units of execution as part of device message processing and the release extends the previous SDK allowing containers to be executed as part of this processing pipeline. You can use either Linux containers for Docker or Windows containers for Docker.
Image source: https://docs.microsoft.com/en-us/azure/iot-edge/how-iot-edge-works
IoT Edge adds several capabilities to the previous SDK including:
- AI Toolkit for Azure IoT Edge
- Azure Machine Learning
- Azure Stream Analytics
- Azure Functions
- Custom modules
The AI Toolkit is a set of scripts, code and tutorials provided to help understand how to apply machine learning at the edge. Azure Stream Analytics uses the same SQL language used in the cloud-based solution, and Azure Functions leverages the same Azure Functions development process that is then containerized for deployment to IoT Edge. By keeping the development processes for each module compatible with the cloud-based version, code and solutions can be reused.
Custom IoT Edge modules can be created in several languages, currently limited to C#, C and Python with more promised in future, and Visual Studio Code can be used to develop, debug and deploy this code in containers for deployment to the edge.
In device scenarios, IoT solutions provide access to a device twin, or device shadow as Amazon in AWS IoT Core call it, which stores the latest device configuration information. In IoT Edge, Microsoft extends this concept with a module twin to provide the same style of cloud-based configuration of the modules that form part of the processing pipeline on the edge device. This can be used to update and reconfigure an edge device or gateway when required as part of normal lifecycle management.
IoT Edge configuration and containers can be deployed from Microsoft Azure IoT Hub and leverage the same security processes.
Many IoT scenarios can benefit from edge computing, for example, image processing to look for defects in manufacturing processes, pre-aggregating sensor data before sending to the cloud or running machine learning algorithms to provide intelligence at the edge.
This is particularly useful where latency or bandwidth is a problem, or where local processing can provide warnings and alerts in cases where an internet connection cannot be guaranteed.
Several customers have used IoT Edge in private preview. In the announcement, Matt Boujonnier, analytics application architect for Schneider Electric, said:
Azure IoT Edge provided an easy way to package and deploy our Machine Learning applications. Traditionally, machine learning is something that has only run in the cloud, but for many IoT scenarios that isn’t good enough, because you want to run your application as close as possible to any events. Now we have the flexibility to run it in the cloud or at the edge.
IoT Edge can run or x64 or ARM hardware architectures and allows Microsoft to compete with other edge computing solutions such as AWS Greengrass, IBM Watson Edge Analytics and SAP Leonardo IoT Edge.