The software giant shares its strategy for getting enterprises on the cloud with advanced analytics, machine learning and support for developer-friendly application containers.
REDMOND, WASH.—Windows 10 just launched, but Microsoft's head already is back in the cloud.
The tech giant gathered industry watchers at the company's sprawling campus here in Redmond, Wash., where company executives demonstrated some of its latest innovations in enterprise cloud software and services. Having already come to grips with the basics of cloud architectures that support enterprise-grade applications, the company is now aligning its analytics, business intelligence and machine learning technologies to provide businesses with next-generation toolsets.
First, Microsoft is growing an expansive cloud portfolio that "powers the apps and services across the devices people love," said Frank Shaw, corporate vice president of communications for Microsoft. Delivering that capability requires a massive cloud, he argued, namely his company's own globe-spanning Azure cloud computing platform, which he described as infinitely scalable, accessible across the world and used by consumers, businesses and governments.
Building and outfitting cloud data centers is not enough. Microsoft's cloud is continually evolving, said Shaw. In the last quarter, the company's cloud teams made "400 additions to Azure," he noted.
The company is also pushing technologies like Azure Machine Learning
, made generally available in February. Once reserved for research labs and well-healed corporations, the technology puts predictive analytics in the hands of practically every IT organization, according to Joseph Sirosh, corporate vice president of Microsoft Azure Machine Learning.
"The democratization of machine learning is what we are doing here at Microsoft," he said. Integral to the effort is the Azure Machine Learning Gallery, a cloud-based collection of APIs and "experiments being built by data scientists all over the world and publicly shared," he said. "Think of this like the Pinterest of machine learning."
Ultimately, businesses are seeking to put their data to work, argued Sirosh. Enterprises want to automate decisions by turning "data into intelligent action," he said. "Not only do you want to know what will happen, but ideally you would want the system to tell you exactly what you should do. You want a recipe for action."
Cortana Analytics Suite
encapsulates that vision. The bundle of cloud analytics services brings "data in from all sorts of sources of data; manages that information; puts them in real big data stores … analyzes it with machine learning … and shows it in Power BI."
Beyond Power BI dashboards, businesses can also enlist the suite's namesake virtual assistant technology, Sirosh added. They can "allow Cortana, the personal assistant, to interact with that," he said.
In a business setting, Cortana understands "the context in which you're working," he said. "It can give you a proactive notification that alerts you about things that are of interest to you, and it's all relevant information. Relevance is incredibly important in a data swamp, and Cortana creates relevance for you."
Finally, Microsoft is courting developers by building a container-friendly cloud
by adding support for technologies like the popular Docker application virtualization platform, according to Mark Russinovich, CTO of Microsoft Azure.
Containers provide developers with agility, enabling them to test and iterate their code rapidly. "The agility [developers] get is because the application is completely self-contained and packaged and deployed on top of a running OS [operating system]," rather than booting a full virtual machine, he explained.
Resource isolation and namespace virtualization also help provide a solid foundation on which to test and ultimately deploy new applications. Russinovich noted that "developers get a very consistent experience with their applications, whether they're debugging it on their own desktop inside of a virtual machine on their desktop or if they're using a Linux client directly on the machine that they're running on."