Microsoft has released a preview of Apache Storm in its Hadoop-on-Azure service.

We reported on Microsoft's plans to add machine learning capabilities to the Azure platform in June 2014 see, Machine Learning Goes Azure - Azure ML Announced. Now we have news of the solutions that will take this initiative further, starting with Storm, which is primarily an analytics tool.

Storm is an open-source app that can be used for real-time analytics of streamed Hadoop data. Customers using Microsoft’s cloud-based version of Hadoop, HDInsight, will now be able to use Storm for real-time analytics. Storm also has machine learning capabilities.

Writing about the new support on the Microsoft blog, T.K. Rengarajan, Corporate Vice President of Microsoft’s Data Platform division, said that Microsoft customers:

“collect and store more data than ever before, expect more from their data and want more insights from it, including being able to do real-time analytics over streams of data to complement their existing Hadoop deployments.”

The preview version of Apache Storm in HDInsight will, according to the blog post “allow our customers to process millions of items of Hadoop data from their Internet of Things devices in near real time using a fully managed Hadoop service.”

Microsoft has also announced plans for hybrid data connectors for moving data to Azure. These will be based on the next version of Hortonworks Data Platform (HDP). Hortonworks HDP 2.2, which includes hybrid connectors to move data to Microsoft Azure, will be available in November.

The machine learning side of the announcement, Microsoft Azure Machine Learning, is also available now in preview. It can be used to design, test, automate and manage predictive analytics solutions in the cloud. The new machine learning capabilities in the Azure Marketplace will let Microsoft customers access machine learning capabilities as Web services.

The services include a recommendation engine for adding product recommendations to a website, an anomaly detection service for predictive maintenance or fraud detection and a set of R packages for programming data analysis systems. All will be available as finished examples for anyone to try. More details are available on the Microsoft machine learning blog.