Microsoft is trying to be the platform company for big data apps, too

The race is on to become the software vendor with the best platform on which to run big data applications, and Microsoft thinks it stands as good a chance as anybody to win. At an event on Tuesday, the company detailed its vision of the future of computing — one that CEO Satya Nadella said entails “ubiquitous computing and ambient intelligence” — and announced three new products it hopes will help its vision come to life.
What Nadella had to say was really a continuation of what Microsoft has been talking about for the past year or so. The company has had to make itself into a data-driven company in order to compete in everything from web search to cloud computing (“Every aspect of Microsoft’s business is being fundamentally transformed because of data,” he said) and it’s going to help its customers undergo the same transformation.
Today, that means faster databases (Microsoft announced SQL Server 2014 with expanded in-memory capabilities), easier SQL queries on Hadoop (Microsoft announced the Analytics Platform Service, which bundles SQL Server and Hadoop on the same cluster with a shared query engine) and an easier way to ingest all that data you’re storing (the company announced a preview of its Azure Intelligent Systems Service for getting sensor data into the cloud). The immediate result: Microsoft will “transform Office as the UI for data,” Nadella explained.
We’ve already seen some of what that will look like with new visualization capabilities and the Q&A feature for Microsoft’s PowerBI Excel add-on, and it looks pretty good. Bigger datasets, faster performance and smarter results.

Counting and visualizing Olympic medals with natural language using Q&A.

Counting and visualizing Olympic medals with natural language using Q&A.

However, this is hardly visionary stuff. If this were all Microsoft had up its sleeve, it would look like just another data management vendor whose line of sight doesn’t go much beyond SQL queries and business intelligence.

The future is in machine data and machine learning

The real future is the one with the ubiquitous computing and the ambient intelligence, the internet of things. And that’s why the Azure Intelligent Systems Service is so interesting. Microsoft, like EMC-VMware spinoff Pivotal (which, not coincidentally is ran by former Microsoft vice president Paul Maritz), wants to become the best place for running next-generation applications, not just analytic workloads. Whether that’s in the cloud, in the data center or both doesn’t really matter.
Already, Microsoft corporate vice president of Quentin Clark told me in an interview after the product launch, Microsoft is working with a number of IoT-like applications industries from automobiles to farming. For example, he explained, one software-as-a-service vendor built an application on Azure that analyzes sensor data from cows in order to help farmers optimize their dairy production. This was all done before the advent of the new service, which should make Azure look like a more compelling platform by making it easier to get data into.
The next step, which is where Microsoft thinks it has a real advantage, is in adding machine learning to the mix. Nadella hinted that machine learning will make its way into the company’s various products in order to deliver the intelligence he mentioned, and Clark confirmed this.

“Microsoft, by nature of what it is … we’ve had to develop machine learning into an engineering practice,” he added.

Whether it’s infrastructural improvements to enable machine learning on Hadoop, like Microsoft Research is working on with its REEF project (which is now open source), or advanced deep learning models for speech recognition (see embedded video below for more on this), the company does have some chops in this space. A platform that encompasses cloud and physical servers, multiple data management frameworks, and machine learning capabilities to help build smart applications sounds good on paper.

[protected-iframe id=”5b1d4c689eacca6ae02657d0c2f287a5-14960843-6578147″ info=”″ width=”640″ height=”360″ frameborder=”0″ scrolling=”no”]

What’s more, Clark noted, Microsoft is “certainly open to” the idea of actually building and selling some turnkey applications should customers demand it. This is a departure from the stance of most Hadoop vendors, such as Cloudera, and even Pivotal (Maritz acquired a handful of application assets while he was CEO of VMware, but VMware sold most of them off rather than send them along to Pivotal), but Microsoft is no stranger to developing the core applications for its platform.

But despite all the money and all the talk, this is still just the beginning of the data revolution in IT. Cloudera, Pivotal, Google, Amazon Web Services, IBM — they all want some or all of the market Microsoft is going after, and they’re all arguably in a position to get it. Microsoft’s strategy seems complete, possibly the most complete, but the company could become a B-list player really fast if it can’t execute.