Microsoft is placing its bets on Artificial intelligence (AI) and hybrid cloud. At Ignite 2017 in Orlando, Redmond emphasized on how AI has become the key ingredient of everything it's developing. Azure Stack, the hybrid cloud offering is available to enterprise customers through select OEM partners.
When Microsoft gets serious about an emerging technology, it starts with the developers. Microsoft is following the same path for making AI accessible. Firstly, it turned AI into a platform that becomes the foundation for both internal applications coming from Microsoft as well as for external developers who can access it through APIs. It is then building a set of tools that make it easy for developers and data scientists to create AI-enabled applications.
Microsoft is committed to embedding AI into almost every new product and service. Microsoft Excel has got new formulae for performing predictive analytics in the cloud. PowerPoint has got a translator that can translate presentations in real time. Word is all set to have a new spell checker and grammar tool that goes beyond the basic correction. But, Office is just one of the products that will have powerful AI features. Dynamics CRM, SQL Server, Bing and many other services will exploit AI capabilities.
SQL Server 2017 is one of the first databases in the industry to get an embedded ML engine. Customers can mix and match existing SQL notations with predictive analytics. The ML engine supports R and Python languages along with modern libraries for training and visualization.
Machine Learning (ML) is at the heart of AI. To enable developers to embrace ML for building intelligent applications, Microsoft has unveiled an ML Workbench, which runs on both Windows and Mac. This tool is targeted at data scientists who are not users of Visual Studio – Microsoft’s flagship development environment. The ML Workbench deals with data cleansing and preparation, which forms the very first step in building ML models. It is integrated with popular open source data science toolkits such as Python Scikit Learn, Jupyter and Matplotlib. The best thing about the workbench is that it integrates with the cloud by seamlessly moving the heavy lifting to the GPU-powered VMs running in Azure.
For developers who are familiar with Visual Studio, there are extensions for popular ML frameworks such as CNTK, TensorFlow and MXNet.
Microsoft has also announced Azure Machine Learning Experimentation service for developers and data scientists to increase their rate of experimentation. Data scientists can use the service for training the models on their local machine, in Docker containers running locally or in the cloud, or on scale-out engines in Azure like Spark on HDInsight. The goal of this tool is to enable rapid iteration for evolving accurate ML models.
The third service in the ML toolchain is the Model Management service that relies on Docker containers to help developers and data scientists to deploy and scale their models anywhere a Docker container can be deployed.
Eventually, when Azure Container Service supports GPUs, the hosted container management platform will become the preferred testbed for model management.
While almost every platform company is talking about utilizing AI, Microsoft is serious about democratizing AI and ML. With access to vast amount of data and customer use cases, the company is expanding the horizon to support new scenarios. These advances and innovations are baked into the platform that is exposed to the developers through APIs and tools.
For Microsoft, AI is the new .NET. I vividly remember the emphasis on embedding .NET platform in every enterprise server that Microsoft shipped back in 2005. The leadership team under Satya has given the same mandate to the product teams.
One of the other key takeaways from Ignite is Azure Stack, Microsoft’s hybrid cloud platform that’s been in the works for a long time. Unlike other offerings from its competitors, Azure Stack runs on certified hardware that is available only through select OEMs. Initially, customers can order Azure Stack from Dell EMC, Lenovo and HP Enterprise. Other hardware vendors will be introduced shortly.
Microsoft positions Azure Stack as an extension of Azure, which delivers a truly consistent hybrid cloud platform.
The consistency between the local and public cloud reduces complexity, which helps customers maximize their investments across cloud and on-premises environments. For developers, consistency enables them to build and deploy applications using the same approach – same APIs, same DevOps tools and the same portal.
Azure Stack supports the core building blocks of Azure – compute, storage, network and identity – delivered through IaaS, PaaS, and management services. Additional services available in the Azure public cloud will be eventually come to Azure Stack, which reduces the gap between the two platforms.
Azure Stack, with 100% compatibility with the public cloud seems to be a viable hybrid cloud platform for enterprises. Can Microsoft convince VMware shops to embrace Azure Stack? Let’s wait and watch.
