Azure IoT principal program manager Arjmand Samuel gave the example of a drone fitted with a video camera and a processor that is capable of detecting anomalies in pipelines.
The Azure IoT approach is to provide a consistent environment for software that can be deployed in the cloud or on edge devices, and to configure devices and provision them with the software required.
Certified hardware identifies itself to Azure based on its TPM, and in return is bootstrapped and provisioned with the modules corresponding to its tags. The Device Provisioning Service is said to make it possible to securely provision tens of thousands of devices.
Azure IoT Edge has been open sourced, and it uses open source components including Moby (which works with Docker-compatible containers).
Modules can be written in C, C#, Node, Python or Java. Organisations can develop their own modules, or buy readymade third-party modules through the Azure Marketplace.
"We designed it [Azure IoT Edge] with security from the ground up," said Azure IoT director Sam George, who explained that Azure IoT Edge guarantees the security of both the software and the edge device.
Early adopters include energy company Chevron.
Chevron fellow and IIoT centre of excellence lead Deon Rae said: "Edge computing is the next wave of cloud and IIoT innovation. With the ability to run Azure IoT Edge on Azure Stack, we see an opportunity to increase uptime and real-time insights on the performance of our operating equipment with intelligent applications that can run right on the device and in remote areas with limited or interrupted connectivity."
Azure IoT Edge is free, but requires an Azure IoT Hub instance for edge device management and deployment. The latter is free for up to 8000 messages per day, after which prices start at US$31.85 a month for up to 400,000 messages.
Azure IoT Edge projects may also use other Azure services (such as Stream Analytics or Machine Learning) which are priced according to use, and third-party modules are likely to come at a price.