Intelligent Enterprise

Wal-Mart, a data warehousing pioneer, is showing that it's not afraid to take a gamble if there's more business insight to be gleaned from the 800 million transactions generated by its 30 million customers each day.

The $345 billion-a-year retailer revealed last week that it's one of the earliest customers of Hewlett-Packard's new Neoview data warehousing system. That Wal-Mart would choose Neoview was unexpected, though not entirely surprising. Unexpected because Neoview is new and unproven, formally introduced just three months ago, and because Wal-Mart is heavily invested in Teradata's data warehousing platform. But it's not surprising because HP CEO Mark Hurd, who once headed Teradata, and HP CIO Randy Mott, who once ran Wal-Mart's IT organization, have firsthand knowledge of the retailer's data warehousing environment and almost certainly drew on that experience to close the deal.

	With 30 million customers a day, that's a lot of toothpaste. Photo by Ruandh Steward

Neoview will be used in conjunction with Wal-Mart's strategically important Retail Link system, which gives Wal-Mart's 20,000 suppliers access to data about the movement and sales of their products in its stores. In other words, the HP platform isn't being relegated to some secondary business process. HP couldn't have hoped for a better proof-of-concept customer as it tries to sell its new business intelligence platform in a highly competitive market that includes IBM, Oracle, SAS Institute, and Teradata.

For more than a decade, Wal-Mart has operated one of the largest commercial data warehouses in the world. Over the last two years, that Teradata-based warehouse has doubled in size to more than 1,000 Tbytes, or a petabyte, packed with sales information on every item sold in its stores. "Business intelligence is huge" at Wal-Mart, says CTO Nancy Stewart. Wal-Mart is making "significant investments" in BI tools, Stewart adds, though she declined to reveal how much it's spending on the HP system.

Earlier this year, Wal-Mart created a loose-knit internal BI team under senior VP Marc Rosen that's assessing new graphics-rich desktop tools that render "what-if" scenarios, as well as new options for cleansing and managing data with an eye on governance. The company also is considering building data marts--smaller, subject-specific data warehouses--that focus on snapshots of operational data, Stewart says.

Wal-Mart has long used its data warehouse and associated BI tools to analyze in-store sales, but there's more to be done in determining the ideal mix of items for each store's customers and to place products in such a way that customers fill their shopping carts. Wal-Mart's aggressive use of radio frequency identification technology--its partners have been forced to adopt RFID to keep up with Wal-Mart--has compounded data growth. "That's a ton of data crunching," Stewart says.