Microsoft Releases Second Preview of Hadoop-Based Service for Windows AzureThe Hadoop-based service for Windows Azure capacity is expanded by four times.

Hadoop Hubbub

In February, we wrote about a leaked slide that revealed the Microsoft Hadoop roadmap:

As you can see, the roadmap indicates Microsoft is planning to simultaneously deliver the final version of Hadoop on Azure and a tech preview of Hadoop on Windows Server on March 30. The final release of the Windows Server offering is planned for June 29."

Then in the first week of March, Microsoft announced plans to release a new, limited preview of the Apache Hadoop-based service for Azure in Q1 of this year, saying:

Since the first limited preview of the service was released in December, customers such as Webtrends and University of Dundee have been testing the service to glean actionable insights from complex data sets hosted in the cloud. As a result, Microsoft is expanding the preview from 400 to nearly 2000 invitation codes."

The company said that the most important new features of the preview include an open source library, Mahout, which helps with building predictive analytics applications, and disaster recovery of the Hadoop Distributed File System (HDFS) NameNode.

Azure Embraces Open Source

At the end of last year, we noted that Microsoft has been attempting to make its Azure cloud service more attractive by improving its usability and showing support for open source development. In addition to releasing the first preview of Hadoop for Azure, the company released an Azure software development kit (SDK) for .Net, Java and Node.js, which is licensed under the Apache license and available on GitHub. Microsoft also updated several open source tools, such as the Azure plugin for Eclipse, tripled the maximum data size in SQL Azure, and added support for SQL Azure Federation.

In a blog post earlier this week, the SQL Server Team announced the general availability of SQL Server 2012, and also slipped in an announcement about the second preview of the Hadoop-based Windows Azure service. The post said that the team expanded capacity of the service by up to four times to accommodate the high demand.

"This preview provides an elastic Hadoop service with more reliability through disaster recovery of the Hadoop Distributed File System (HDFS) NameNode, and support for advanced analytics through Hadoop projects like Mahout," the blog post explains.

Customers can sign up for the latest preview on the Hadoop on Azure website.