Netezza’s Developer Network looks to draw developers, systems integrators, academia, and complementary ISV partners into a global, tight-knit community that advances NPS’ patented non-SQL streaming architecture called Intelligent Query Streaming. The architecture embeds complex analytic algorithms within the core NPS appliance platform, allowing them to be run on-stream – i.e. as fast as the data streams off the disk.
Developer Network members get access to a small development environment called a SPU-box that emulates Netezza’s Snippet Processing Units (SPUs) which are massively parallel blades responsible for processing individual query segments on a portion of datasets, and an SDK. Because the SPUs scale upwards in a linear fashion, developers can prototype small analytic applications safe in the knowledge that they will map directly to larger-scale deployments by simply adding more SPU nodes. For example a developer can create a half-terabyte application in on their desktop using the SPU-box and roll it out into production as an 8-rack server system.
Netezza has already signed up a handful of founding members, including 10e Solutions, Intelligent Integration Systems, SAS Institute, SPSS, and Carnegie Mellon University.
Netezza said these members are already working on range of algorithms such as multidimensional spatial analytics for risk management, iterative modeling of call records for telco price optimization, predictive model scoring for customer segmentation, and fuzzy text search for best guess results.
According to Ellen Rubin, vice president of marketing at Netezza, the Developer Network signals a shift in NPS’ positioning away from just data warehouse appliance to more broadly applicable analytic appliance.
We want to advance the role of data warehouse appliances. As we move processing as close to where the data is physically stored that opens up a much broader range of analytics beyond traditional BI reporting and analysis.
Rubin said that Netezza will refer to its appliances in the context of analytic, as opposed to just data warehousing, environments from now on.
What we’re doing with our Developer Network is to pull custom, complex analytic code and algorithms developed by third-parties into our appliance.
Justin Lindsey, Netezza’s CTO, said the move to embrace deeper and more complex analytics using an on-stream approach is a natural evolution for Netezza’s appliance platform.
The initial value for NPS was fast processing of queries on our custom hardware design and leaving as much data on disk. People got comfortable with that inside our machine but found it painful to move it outside for custom processing for more complex analytic processing.
By complex analytic processing he means applications like predictive analysis for re-pricing or customer scoring, image analysis or complex hashing algorithms that deal with different types of non-SQL data like geospatial information.
What we’re doing now is putting those same algorithms inside the machine instead of having users to move them off at a distance and pay a performance penalty for that.
We’ve already built a scalable platform to enable that processing without moving out of the data warehousing and into custom machines. But it’s important to connect to more powerful algorithms. We’re doing that with our Developer Network partner model.
Netezza is based in Framingham, Massachusetts, and this summer became the first data warehouse appliance specialist to IPO on NYSE. The company has around 110 customers to date.
The company’s user conference is running this week in Boston and drew in a crowd of around 500 attendees.
