Microsoft Gives Details on Its Analytics Platform System

By Pedro Hernandez | Posted 2014-04-26 Email Print

By combining SQL Server and Hadoop, Microsoft is positioning APS as an affordable gateway to big data analytics for enterprises.

Last week, during an April 15 Internet of things-themed customer event, Microsoft officially released Analytics Platform System (APS), a SQL Server-based offering that leverages Hadoop that the company asserts changes the economics of gleaning business-boosting insights from massive stores of data. The product arrived amid the backdrop of SQL Server 2014's launch. The company envisions SQL Server as a foundational software element of the Internet of things. As such, the latest version of the database includes enhanced capabilities like in-memory online transaction processing (OLTP) to speed up the processing of data from a flood of Internet-enabled devices, mobile users and business systems. Apache Hadoop, the popular open-source batch-processing platform, has emerged as the dominant software foundation for big data analytics. Microsoft's Hadoop distribution is based on the Hortonworks Data Platform. Combined in APS, SQL and Hadoop enables businesses to query Hadoop data without requiring IT to preload it into the data warehouse and quickly explore that data courtesy of Microsoft business intelligence (BI) integration. The company claims huge performance gains (up to 100 times) over legacy data warehouses.

Now Microsoft is spilling more details about APS and how it serves as a turnkey appliance for enterprises seeking to glean BI from big data.

'APS is an evolution of our SQL Server Parallel Data Warehouse (PDW) appliance, which builds upon the high-performance and scale capabilities of that MPP [massively parallel processing] version of SQL Server,' stated the company in a blog post. The product, added Microsoft, 'now introduces a dedicated region to the appliance for Hadoop in addition to the SQL Server PDW [Parallel Data Warehouse] capabilities.' PDW is Microsoft's shared nothing, parallel processing architecture that speeds complex queries and delivers improved performance while simultaneously loading and querying data, compared with traditional methods. Based on SQL Server, the software can scale linearly froma few terabytes to 6 petabytes, according to the company. Hardware partners include Dell, HP and Quanta Computer, each of which offers converged, pre-integrated Microsoft APS hardware solutions. Microsoft explained that the APS appliance's Hadoop region includes several add-ons. Key capabilities include 'high availability through the appliance design and Windows Server failover clustering, security through Active Directory and a unified appliance management experience through Systems Center.'

Further integration was made possible by a team in Microsoft's Grey Systems Labs, a design and development group within the SQL Server division. 'Completing the APS package and seamlessly unifying the data in SQL Server PDW with data in Hadoop is PolyBase, a ground-breaking query technology developed by Dr. David DeWitt and his team,' said the company. Given the technology's roots, APS can also potentially benefit SQL pros as demand grows for big data solutions in the enterprise. In discussing SQL 2014 and how it empowers database administrators (DBAs) to bridge IT and business, Denise McInerney, a data engineer at Intuit, said in a another blog post that as businesses gather and store more structured, unstructured and streaming data, DBAs will increasingly face a hybrid data management environment. Echoing the capabilities of Microsoft APS, McInerney said such an environment 'will include SQL Server relational stores integrated with big data solutions such as Hadoop for storing and processing large data sets.'