It seems Hadoop makes for strange bedfellows. At the Microsoft PASS Summit 2011 it was announced that the upcoming versions of SQL Server (2012) and Azure, the Microsoft Cloud service will both support Hadoop. Both of these should be available in preview soon with Azure by the end of the year and the new SQL server sometime in early 2012.
Porting Hadoop to run on Windows instead of Linux which it was designed to run on will be no easy task. Part of the core functionality of Hadoop is that the OS is light and doesn't take a lot of resources up. Few people use that description for Windows. But Microsoft has signed up for the job. Also they have announced that they have engaged Hortonworks to help them with the coding. They have also said that code developed will be donated back to the community.
Having Hortonworks, the spin off consisting of many of the core Yahoo Hadoop team working on this with Microsoft is comforting on several levels. First of all it gives Microsoft some deep subject matter expertise to draw upon. But perhaps more importantly, with Hortonworks commitment to open source and the Apache Foundation it should help allay any fears of the usual suspects who look warily at anything Microsoft does regarding open source.
What many expect is what Microsoft will offer is a set of connectors between MS SQL and Hadoop that will allow you to move data back and forth between the two. But running Hadoop on Windows may be the most difficult task on this menu.
Part of the plan is also to have other Microsoft tools access data in Hadoop. Can you imagine importing Hadoop data into Excel for instance? You have to give Microsoft credit here for the vision and the apparent good intentions in being a good open source player here.
Microsoft supporting open source projects, donating code back and promising compatibility with the Apache version of Hadoop? What will be next, yellow elephants flying?