IBM, SPSS Partner to Embed Analytics in Data WarehousesIBM, SPSS Partner to Embed Analytics in Data Warehouses

IBM announced October 31 that it is working with analytics vendor SPSS to deliver predictive analytics capabilities within IBM DB2 Warehouse 9.5, enabling customers to handle data mining and scoring in real time inside the data warehouse, rather than as a separate, time-consuming offline batch-oriented task.

The IBM-SPSS tie follows closely on the heels of a similar announcement by SAS and Teradata in early October that the partners will embed SAS analytic models within the Teradata database. Nonetheless, IBM claims it has been leading, not following, in the drive to embed analytics in the database.

"We were able to run SPSS services inside our warehouse more than a year ago, but that wasn't with version 9.5 with its Extreme Workload Management capabilties," says Marc Andrews, IBM's Program Director of Data Warehousing. "We can now run SPSS models and manage the workload so we can ensure the appropriate prioritization of analytic workloads."

DB2 Warehouse 9.5 was unveiled two weeks ago at IBM's Information On Demand conference. The new Extreme Workload Management feature assigns warehouse system resources based on processing priorities. In a call center application involving predictive churn analysis, for example, the processing must be handled in seconds while the customer is still on the phone. At the other extreme, routine backoffice queries and reports can wait.

SPSS’s Clementine predictive analytics technology and Predictive Enterprise Services platform will be integrated with IBM's DB2 Warehouse 9.5, providing a graphical interface for fast creation and use of data mining models to uncover and visualize data patterns and apply this insight to improve business processes. SPSS's technology will also automate the storage and management of predictive analytic tasks that apply predictions to business systems.

Data warehouse appliance vendor Netezza has also chimed in on the embedded-analytics trend, opening up the code to its Netezza Performance Server, releasing a software developers' kit and forming a Netezza Developer Network with SAS and SPSS counted among more than 30 members.

Teradata and Netezza both point to processing-speed advantages tied to their hardware-based, shared-nothing architectures, and IBM now claims similar advantages. "We've had a shared-nothing architecture in our software for a while, but the "Balanced Warehouse" products we introduced in March offer an appliance-based approach that combines hardware, software and storage in a massively parallel processing architecture," says Andrews.