HPE Expands Machine Learning Capabilities for Big Data Analytics - Page 2

The company already has 18,000 developers working with Haven OnDemand—part of HPE's larger effort to provide machine learning capabilities to a wider group of users—and Combinations is available now for early access, with general availability coming in the fourth quarter.

Key new features in Vertica 8—which had been code-named Frontloader—include support for more cloud platforms, in-database capabilities and enhanced capabilities to access and analyze data residing in multiple places. The features help address customer demands for greater openness and flexibility when it comes to the huge amounts of data being generated and stored, according to Colin Mahony, senior vice president and general manager of HPE Software's Big Data Platform business.

"When it comes to choice, customers want to run [their analytics workloads] on a lot of different platforms," Mahony said at the show.

That includes a choice of cloud platforms, according to officials. Vertica, which already supports Amazon Web Services (AWS), can now work with Azure, a move that not only increases the options for customers but also builds on a strategic agreement HPE and Microsoft forged late last year to work together on hybrid cloud solutions.

In addition, through Vertica 8—which will be released in the fourth quarter—users can now more easily and securely access and analyze data that resides in Hadoop data lakes. Until now, businesses could put Vertica into Hadoop, which enabled them to analyze that data with Hadoop-like economics but at a cost of performance, Veis said. Or they could use Vertica to access the data, but that meant that the data would have to be copied and moved into Vertica. Now, with new Parquet and ORC readers, customers can use Vertica 8 on data residing in the Hadoop data lake without having to copy and move the data.

"We're bringing the analytics to the data rather than bringing the data to the analytics," Veis said.

In-database machine learning algorithms enable developers to natively create and deploy R-based machine learning models directly in Vertica for large data sets, and a new optimized Apache Spark adapter offers fast data exchange between Vertica and Spark systems.

In addition, core data movement and orchestration improvements can result in 700 percent faster data loading for hundreds of thousands of columns, officials said.