Microsoft Corp. today rolled out a set of new capabilities to Azure that will make it easier for companies to run artificial intelligence software in hybrid cloud environments.

The capabilities are based in part on three existing Azure services. The services are Azure Arc, Azure Machine Learning and the Azure Kubernetes Service.

Azure Arc is an important component of Microsoft’s hybrid cloud strategy. Introduced in 2019, the service enables administrators to manage on-premises servers through the same interface that they use to orchestrate infrastructure resources in Microsoft’s public cloud. For administrators, performing infrastructure maintenance tasks in a single, centralized interface can be more efficient than the traditional approach of using multiple tools. 

Microsoft’s new hybrid cloud AI capabilities also draw on Azure Machine Learning, a set of tools designed to ease neural network development. Another component of Microsoft’s AI feature set for hybrid cloud environments is Azure Kubernetes Service. The service simplifies the task of maintaining software container environments.

Thanks to the new capabilities, Azure customers can now more easily use their on-premises Kubernetes clusters to run AI workloads. A company could leverage its on-premises Kubernetes environment to train an AI model created with Microsoft’s cloud-based Azure Machine Learning toolkit. It’s also possible to perform inference, or the task of running a neural network in production after training is complete.

According to Microsoft, its hybrid cloud AI capabilities support multiple use cases. A company could run AI workloads on-premises to make better use of existing investments in data center infrastructure. Meanwhile, organizations that are adopting a multicloud approach can distribute AI workloads across multiple public cloud platforms.

There are cases where regulatory requirements make it necessary for a company to run important AI applications on on-premises infrastructure. According to Microsoft, its new capabilities can simplify that use case as well by easing common infrastructure management tasks.

After a company connects an on-premises Kubernetes cluster to its Azure machine learning environment, developers can access the cluster through a tool called Azure ML Studio. According to Microsoft, developers are gaining the ability to more easily deploy machine learning workloads on not only servers running in on-premises data centers but also edge computing devices. 

“Now you can build, train and deploy your machine learning models right where the data lives, such as your new or existing hardware and IoT devices,” Kathleen Mitford, the corporate vice president of Azure marketing, wrote in a blog post today. “Azure Arc provides a consistent development, operations and security model for both new and existing applications.”

Azure Arc is an important component of Microsoft’s broader hybrid cloud portfolio, which also includes a number of other offerings. The company sells appliances that enable enterprises to deploy Azure services in their on-premises data centers. Additionally, Microsoft provides an array of networking offerings that an organization can use to link its on-premises systems to Azure.

Photo: Microsoft