Until now, Home Depot has tapped into its mainframe-based operational systems for business intelligence, laboriously extracting data and creating tables for specific analysis projects, says Kevin Murphy, the retailer's information management VP overseeing the warehouse project. DeRodes, CIO at Delta Airlines before taking the Home Depot CIO post in February, acknowledges that the company lags other big companies in data warehousing.
"It's pretty unusual for a company our size to be just starting something like this," he says. "But it becomes an advantage. We're not going to waste a lot of time learning." Warehousing software and practices have matured, he notes, and IT workers with data warehouse experience are plentiful. This isn't the first time Home Depot has been cautious with IT: The company only began selling products through its Web site in late 2000 after carefully considering its E-business options, long after many retailers went online.
The warehouse initially will consist of two 32-node pSeries 690 IBM servers running AIX 5.1 with 60 terabytes of storage. It'll be built on IBM's DB2 database, but Murphy is still talking with a number of vendors about what data-integration and business-intelligence software to use. Initially, the system will have 5 to 10 terabytes of raw data, but neither Murphy nor DeRodes would guess how big it will become.
The data warehouse is slated to go into production by the end of November with a human-resource spending-analysis app. Sales data from Home Depot's 1,500-plus stores will be loaded in the first quarter of next year. Eventually, the warehouse is expected to support apps for sales-forecasting, pricing, product- and store-assortment, space-planning, inventory-management, and purchasing analysis. The system would also feed information to a new "management dashboard" that would display performance data.
The company plans to make the warehouse available to a broad range of top execs, midlevel and store managers, and business analysts.