Doug Miller, Director Database Development & Operations

Kara Cummings

R.L Polk & Co. has served the automotive industry since 1922, providing a variety of analytical and statistical data services. The company collects and compiles data from more than 240 different sources, including state agencies, automotive manufacturers, financing companies, and a variety of providers of lifestyle and demographic data. R.L. Polk’s customers require fast access to this actionable data to make important sales, marketing, and planning decisions. R.L. Polk maintains a four-terabyte data warehouse with data on 500 million unique vehicles and 250 million unique households. As the automotive industry continues to grow, the company has witnessed an exponential growth in the creation of data--making it ever more challenging to maintain the high performance levels of its large databases. R.L. Polk realized it needed to make a change to preserve its competitive advantage in the marketplace, amid significant industry, regulatory, and technology change. Previously, R.L. Polk ran its databases on a mainframe, which was very reliable but did not allow the company to deliver data in real-time to its customers.

The company worked with its wholly-owned subsidiary, RLPTechnologies, and Capgemini to develop the new system. R.L. Polk had a lot of experience with large, symmetric multiprocessing (SMP) servers, which had always proved reliable. However they were very expensive and only allowed R.L Polk to deliver data in batches. R.L. Polk implemented Oracle Database 10g and Oracle Real Application Clusters 10g on four Dell PowerEdge 6850 servers with Intel Xeon dual-core processors running on Red Hat Enterprise Linux. The company is running two databases across the cluster; each has four instances. R.L. Polk had been providing data in batches to its customers for years. By implementing Oracle Database 10g and Oracle RAC 10g, the company is now able to deliver data in real-time. R.L. Polk has a tiered warehouse strategy, in which the Oracle system drives the company’s core data engine, which then feeds data downstream to other databases supporting various applications. The core data engine houses data on vehicle sales, manufacturing, purchases, registration, and other demographic information. It grows in size by approximately 20 million new vehicles each year. With the grid computing architecture from Oracle, R.L. Polk can take a piece of raw data the moment it enters the company and process it to make it available in a data warehouse or data mart, rather than wait to build a critical mass of data that is released in a batch. Tests to date show improvements of up to 70% in data-file processing speed. Further, the system can process 42 transactions per second, nearly twice as fast as R.L. Polk’s design goal of 25 transactions per second. At least 40 of the 50 applications R.L. Polk has in the marketplace run on Oracle data warehouses. At this time, internal facing applications are primarily the ones running on the new Oracle 10g environment, but R.L. Polk is moving toward putting more customer-facing applications on this platform too.