- Platforms: Machine Learning as a Service (MLaaS)
- Azure Machine Learning service
- Bot Framework and Azure Bot Service
- Excel - Data Analysis
- Visual Studio Code
- Project Cortex
- AI School
- Git - GitHub and GitLab
- Microsoft Internet of Things (IoT)
- TensorWatch AI debugging and visualization tool
- Graphical Tools for Modeling AI Components
SQL Server Machine Learning Services
Azure IoT Edge
- Bringing AI to the Edge | Mircosoft
- 5 Reasons Why Azure IoT Edge Is Industry's Most Promising Edge Computing Platform | Janakiram & Associates
- Open sourcing the platform - Azure IoT Edge is available as an open source project on Github. Edge computing use cases are still evolving where customers are expected to use the platform in unique ways. To enable flexibility and openness, Microsoft has opened up the source code of its edge computing platform. Through this, customers will be able to customize their deployments based on Azure IoT Edge. Adding legacy protocols, integrating with existing asset management solutions, interoperability with proprietary communication protocols and data formats becomes possible through the customization of Azure IoT Edge source code. Open sourcing edge is a brilliant move from Microsoft. It only increases the trust and confidence of customers.
- Containers at the core - Microsoft has adopted Moby, the open source container runtime that powers Docker, as the engine for Azure IoT Edge. This design decision enables developers to package and deploy standard Docker containers as modules on Azure IoT Edge. Similar to UNIX Pipes, the output of one module can be fed as an input to another module creating a logical chain of Docker containers that work in tandem. Microsoft is making some of the Azure services such as Stream Analytics, Functions, and SQL Server as containerized modules for the edge. Each module can be managed and maintained separately without having to deploy the entire application. The container images are stored in the standard registry either in the cloud or within the data center. Customers can build CI/CD pipelines to automatically push the latest version of modules (container images) to multiple edge locations. Microsoft is also exploring the integration of Kubernetes with Azure IoT to orchestrate the distributed edge deployments effectively. Adopting containers for packaging both Azure services and custom logic goes a long way in managing complex, distributed edge deployments.
- Ecosystem engagement - Azure IoT already has a vibrant ecosystem of OEMs and ODMs which is now getting extended to Azure IoT Edge. The IoT Edge certification program has the capability-based certification concept. Each vendor participating in the certification program is assigned a level to identify the capability. For example, a vendor targeting the core runtime will get level 1 while another vendor with robust security offering is eligible for level 4. This capability-based marketplace enables customers to choose from a broad ecosystem of partners offering edge computing solutions. Microsoft has also integrated Azure IoT Edge with Visual Studio Team System and Visual Studio Code. Developers can use pre-defined templates to start building the modules. With VSTS, customers can implement CI/CD pipelines to manage the complete lifecycle of modules.
- Security - Azure IoT Edge is a logical extension of Azure IoT platform. It takes advantage of services such as Device Provisioning Service to provision tens of thousands of devices securely. The built-in Security Manager acts as a well-bounded security core for protecting the IoT Edge device and all its components by abstracting the secure silicon hardware. ODMs can choose to harden the platform through Hardware Security Modules (HSM).
- AI @ Edge - Microsoft has made it easy to run machine learning models at the edge. Each model responsible for inferencing can be packaged and deployed as a standard module. Developers can train their models on Azure through Data Science VMs or Azure ML Studio. Azure IoT Edge also supports running models exported from Azure’s Automated Machine Learning (AML) - AutoML services such as custom vision. Since each model is just a container/module, new models can be quickly pushed to the edge. With Microsoft’s investment in ONNX, ML models built using different frameworks may be exported to a standard format before using them for inference. Azure IoT Edge plays a crucial role in Microsoft’s vision of delivering Intelligent Cloud and Intelligent Edge. Some of the design decisions such as containerized modules, tight integration with HSM, plugins for Visual Studio turn Azure IoT Edge into one of the most comprehensive edge computing platforms in the industry.