Netezza opens box, builds developer network for predictive analytics and high-demand, non-SQL analyses.
By Doug Henschen
Appliances are changing the game in data warehousing, and if Netezza has its way, its appliance will also bring big changes to predictive analytics and other forms of high-demand analytic and algorithmic analyses.
Broadening the appeal of its Netezza Performance Server (NPS), the vendor announced this week that it has extended the capabilities of the appliance by opening it up to non-SQL programming, with leading examples including predictive modeling, scoring for marketing segmentation, Monte Carlo simulation, image analysis, geospatial analysis and fuzzy text search.
“Many of these analyses are being done on supercomputers or big grids because of the complexity and performance requirements,” says Justin Lindsey, chief technology officer. “Our appliance is really a high-performance, parallel computing environment, so instead of letting developers just push SQL toward the data, we’re letting them push complex, programmatic functions.” These functions are typically written in fast languages such as C or C++, says Lindsey, but the programming could also be in Java or another language.
Thus, with a new NPS (or a field upgrade of an existing device), customers will be able to handle both warehousing/mart and high-demand analytic/algorithmic activities on a single appliance. And as in warehousing applications, the benefit will be faster processing.
“It took Catalina Marketing about seven hours to handle a particular marketing scoring exercise across billions of records, but by moving it into NPS, they’re going to be able to do it in about half an hour,” says Lindsey.
To accelerate its push into high-end analytics, the vendor has made available a scaled-down development server, a software development kit and partner training. It has also organized and announced a Netezza Developer Network open to customers, systems integrators, complementary technology partners and the academic community. Founding members of the network include SAS, SPSS, Epsilon, Catalina Marketing and Carnegie Mellon University, among others.
“We have granted about 30 partner organizations small development environments so they can open the appliance and do their own programming inside it,” says Ellen Rubin, vice president of marketing. “We’re using the developer network both to drive requirements and to stimulate creativity and innovation.”