Intelligent Enterprise

Faster analysis means faster decisions, so there's a growing trend toward moving calculations and queries into the data warehouse, avoiding data movement that slows response times. Some 18 months ago, SAS and Teradata announced plans to support so-called "in-database analysis," and this week at the SAS Global Forum in Washington, D.C., the two vendors announced the resulting technology and service offerings.

"We now have a joint center of excellence that can help customers from deployment and development though to maintenance and services," says Rob Berman, a vice president at Teradata. "It's unique among partnerships in that it's delivered as one offering, from taking the order through delivering the services and ongoing support."

The initial joint technology and service offerings can be delivered through either vendor, depending on the customer's purchasing, deployment and support preferences. Analytic Advantage packages include prepackaged hardware, software and services aimed at streamlining deployments in addition to all the steps leading up to and including actual analytic queries.

"At least 70 percent of the typical analytic lifecycle is tied up just in trying to get the data into shape for creating models," says Scott Van Valkenburgh, director of platforms and ISVs at SAS. "The Teradata data set generator for SAS saves 30 percent to 40 percent of the time normally required to get data ready. The second major time drain is coding and validating models, but our SAS Scoring Accelerator for Teradata instantly makes SAS code run inside Teradata. Finally, once you're processing complex queries inside the Teradata [parallel processing] environment, it's like running all 26 miles of a marathon at the same time."

In-database processing is said to enable real-time applications such as credit scoring while a customer is handling a transaction on an ATM or examining up-to-the-minute profitability or risk while a customer is still on the phone with a customer service rep.

Analytic Advantage packages are paired with scale-appropriate Teradata data warehouses aimed at three levels of practitioners. An Express package targets entry-level and departmental practitioners and includes the SAS Analytics Pro analytic toolset, a prebuilt SAS/Access Interface to Teradata's relational database and a SAS Enterprise Guide Windows graphical user interface. An Advanced package is aimed at experienced users and includes the Express components plus SAS Enterprise Miner data mining software and the SAS Scoring Accelerator for Teradata, which quickly translates SAS models into Teradata-specific functions for scoring. An Enterprise package adds the SAS Model Manager, which streamlines the error-prone steps of creating, managing and deploying analytical models. This offering is geared to large deployments where many models are created and revised across the enterprise.

Adjustments and upgrades will be offered to customers who already have components of SAS or Teradata technology. Similarly, Optimization Services packages can be tailored to the specific training needs of the customer. Given the variability and customized nature of deployments, package costs were not detailed.

Several joint customers in the retail, financial services, insurance and travel industries are said to have purchased Analytic Advantage packages, according to SAS and Teradata, but none was prepared to discuss its deployment. SAS and Teradata currently have approximately 400 joint customers worldwide.

SAS and Teradata aren't alone in pursuing the speed advantages of executing analytic models from within a database. As reported in this article, data warehouse vendors Netezza, Greenplum and Aster Data Systems are also working with software vendors and developers, and at least two customers of Netezza's "on-stream analytics" offering have been quoted about their deployments.

The Analytics Advantage and Optimization Services packages are billed as the first two offerings in a coming wave from SAS and Teradata. Anti-money-laundering, credit risk/credit scoring and Enterprise Intelligence packages are said to be in the works.