News and events

Kognitio announces Massively Parallel R via External Scripts for Computationally Intensive Processes

Mar 19, 2013 | By admin

Data Science Lab Functionality Delivered With Enhanced Capability; A Principal Aspect Of The “Logical Data Warehouse”

Grapevine, Texas, March 19, 2013 — Kognitio, driving the convergence of Big Data, in-memory analytics and cloud computing, today announced the Kognitio Analytical Platform enables new fully parallel not-only-SQL (NoSQL) capabilities, most notably the R language for statistical computing and graphics.

The announcement came at the Gartner Business Intelligence and Analytics Summit 2013 in Grapevine, Texas, where Kognitio is demonstrating the technology this week. The company is also sponsoring an informative session with case studies on the growth of “Big Data” and how the Kognitio Analytical Platform has helped clients in media, retail and customer loyalty analytics.

The capability of parallelizing R is part of the groundbreaking external scripting functionality in the latest release of the Kognitio Analytical Platform. External Scripting enables a user to invoke any programming language that can run under Linux from within the Kognitio platform environment. Leading analysts cite this capability as a principal aspect of what they term the “logical data warehouse;” they say it provides a data science lab functionality that enables distributed processing, which is an emerging best practice for analytical data management.

Computationally intense processes like product forecasting, Monte Carlo scoring and robust clustering have proven challenging over rapidly growing volumes of data. The demand by businesses for ever-faster results can now be delivered via the Kognitio Analytical Platform in near real-time, as opposed to the hours or days required by competing solutions.

“Our internal testing has shown statistical algorithms executed in R run thousands of times faster due to the ability to simply parallelize standard scripts with industry-standard SQL managing data flows and aggregations,” said Paul Groom, Kognitio chief innovation officer. “Data sets with billions of rows are automatically partitioned and fed to thousands of parallel R instances with results flowing back into standard SQL constructs.”

Groom noted that data scientists can design R jobs as they normally would, using the Kognitio Analytical Platform to parallelize and run thousands of jobs quickly from a single query. With the data already resident in-memory, they can invoke massively parallel processing from within the analytical platform environment – affording them the substantial benefit of running it in close proximity to data persistence/storage. The Kognitio external tables functionality allows data to be brought directly into RAM from a range of sources such as Hadoop.

The company said it has spoken with several large retailers about the added NoSQL capabilities who believe it will provide them breakout capabilities not seen in other solutions, providing them with a strategic advantage over their competitors.

Kognitio has been consistently at the forefront of advanced analytics, leveraging its generation of experience as an in-memory analytical platform to enable companies of all sizes worldwide to better understand the insights available from massive amounts of data, so that they can gain and maintain a competitive advantage in the marketplace.

Executives noted that this capability will be fully available for production environments in the Kognitio version 8 release of the software that is generally available in June. However, users can try it today for free by contacting them or visiting their web site.