Starting today, Azure IoT Edge for Linux on Windows (EFLOW) allows you to run cloud native Linux workloads on Windows IoT in production deployments. You can now leverage the best of both platforms rather than choosing between Windows and Linux for your IoT solutions.
Enterprises have told us that they want to take advantage of the large number of Linux-based cloud-native containerized workloads on the edge, especially for AI/ML. Many of those same customers also value the unique benefits of Windows IoT, such as the ability to create interactive user interfaces with natural input, enterprise grade device management tools, world-class security, 10-years of long-term servicing from kernel to shell, and a worldwide update service. Yet deploying, managing, and maintaining both Linux and Windows IoT devices is time consuming and expensive. EFLOW allows Windows IoT customers to retain their existing devices, tooling, and application investments, while also running Linux containers on the same devices. Conversely, enterprises with existing Linux investments who are interested in the benefits of Windows IoT have a migration path to bring their Linux application code to Windows.
EFLOW is a lightweight Linux VM from Microsoft designed for production edge deployments. It is based CBL-Mariner, a Linux distribution developed and maintained by Microsoft. It includes Azure IoT Edge to facilitate easy integration with the cloud and deployment of workloads from Azure IoT Hub to Windows devices on the edge. It also supports access to commonly used embedded hardware, including TPM, serial, and Nvidia T4 and GeForce/QuadroGPUs for AI/ML acceleration. We plan to extend support to Intel iGPUs for AI/ML use cases by the end of the year. Furthermore, EFLOW supports deployment of the VisionOnEdge (VoE) solution template, which illustrates how customers can create their own AI/ML solution, using third party or Azure technologies, such as Azure Video Analyzer. As a developer, you can use VoE as a starting point or you may choose to implement your own custom modules using the Linux distribution of your choice. You can develop Linux workloads using the Windows Subsystem for Linux (WSL), which is based on the same Linux kernel as EFLOW. Alternatively, the Azure Marketplace from Microsoft offers prebuilt 1P solutions, such as SQL Edge and OPC Publisher, as well as 3P modules from our partners (eg. OpenVino) that can be deployed as-is. Finally, just like Windows itself, Microsoft updates the EFLOW VM with the latest critical fixes and security patches using the same global servicing infrastructure used for Office and Windows.
Windows IoT is deployed in millions of embedded devices around the world in numerous industries including manufacturing, retail, medical equipment, and public safety. Customers choose Windows to power their edge operations because it is an out-of-the-box platform to create locked-down, interactive user experiences with natural input, provides world class security, enterprise grade device management, and 10 years of servicing, allowing customers to build solutions that are designed to last. With EFLOW, customers will now be able to benefit from running Linux workloads on production Windows IoT deployments, leverage the advances in cloud-native development, and easily connect the solution to Azure.
EFLOW is available on all Hyper-V capable Windows 10 2019 or newer operating systems (build 17763+), including IoT Enterprise, Pro, Server, and Enterprise. This makes 100s of millions of existing devices EFLOW capable, which can easily be managed and connected through Azure.
Learn more about EFLOW by watching the IoT show:
Detailed documentation to get started is available at https://aka.ms/AzIoTEdgeforLinuxOnWindows
If you want to stay up to date and get notified of future updates to Azure IoT Edge for Linux on Windows, you can register using this link. Note that the information you will share will only be used by Microsoft for the purpose of keeping you informed about this product.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.