≡ Menu

First Look – Revolution and Netezza


Revolution Analytics and Netezza have recently announced a partnership. Netezza has a strong history of being associated with customers that are focused on “big data” problems and advanced analytics. As a result they have been delivering in-database analytics to bring these analytics closer to the data. Their approach allows customers and partners to bring Java, Python,C/C++ Fortan or whatever into the database to execute right by the data and has been used by a number of analytic companies (including SAS and Angoss) to create in-database analytic offerings. The popularity of R has been growing in the Netezza customer base, as it has almost everywhere. This demand is coming from new companies with no prior predictive analytic capabilities, established analytic users who are adding R to their toolkit and some long term R users looking for a more robust, commercial ready platform. To meet this demand Netezza have partnered with Revolution Analytic to port their commercial distribution of R to the Netezza APIs. The companies plan to have it released by the end of the summer.

The offering will support both the creation of models in-database and the execution of the models in database. Thus routines that build models will be able to execute in-database while a model is being developed and the model itself will be able to execute in-database, allowing it to be calculated live when it is needed as part of a database query. As always the benefit of in-database analytics is to eliminate the data movement time.

Once the product is released, companies will be able to use R packages and run them in the Netezza TwinFin appliance. R packages can be applied to a number of data slices in parallel and then aggregated up once they are complete. In addition, Netezza provides a collection of analytic routines that are explicitly parallelized to take full advantage of the distributed computing architecture in the Netezza appliance, and operate on the entire dataset.

Demand is coming from all verticals and, at least in terms of early adopters, is evenly split between companies that want to embed predictive analytics into reports and dashboards (predictive reporting) and those who want to embed in operational systems (decision management). While Revolution Analytics sees more overall demand for predictive reporting there is a higher percentage of those doing decision management that needs the kind of performance that drives in-database demand.

You can get more details on how the combined solution will work here.


Comments on this entry are closed.