San Diego -- “Big Data,” and what to do with it, was a recurring topic at the Teradata Partners User Group Conference, held here from Oct. 2 to 5. Retailers, insurance companies, telecom firms, banks and others looking for best practices in data warehousing and optimization were among the 3,500 attendees at the San Diego Convention Center. With so many challenges facing IT professionals --everything from geomarketing to capturing tweets inside a data warehouse -- participants focused on collaboration and learning from each other’s mistakes and successes.
In a session called Mastering Metadata Management, Kathryn DioQuino of Lowe’s explained how her team built bridges between various departments in order to implement a program that would give easy, shared access to the data warehouse. Her 12-person team, which included members from data, research and analytics, had 12 weeks to accomplish their mission. They succeeded, with some back-up from the infrastructure and network security team.
To hear DioQuino explain it, they did the project on a shoestring. “We hated to use IT resources,” she said. Lowe’s already owned MDS software so the company only needed to pay for some Teradata services. Although the project was run by “data scientists,” as they jokingly renamed themselves, the push came from Lowe’s business side.
The payoff, DioQuino said, is that the company now has data stored from multiple sources in one place. And Lowe’s has much more flexibility in searching and finding customer data, opening up new business intelligence opportunities.
Home Depot representatives led several sessions, including one about data warehouse management and another involving sales lost to stock-outs because there’s not enough inventory on the shelves. A number of attendees checked back in with Clay Barrineau of Home Depot, who gave a session at last year’s conference on Home Depot’s ongoing rollout of its new mobile devices.
“Store Walk Mobility” now goes by a sexier name -- “Tactical Reporting” -- and all 2,000 of Home Depot’s U.S. stores have 15 to 20 of the handheld devices. Stores associates can check inventory for a particular item, or see if a neighboring store has one in stock. These are just two of the types of questions that flow into Home Depot’s data warehouse, which fields 300 to 400 distinct queries from the devices on a typical Monday.
Barrineau and his team have to make sure the system delivers a timely response without riding roughshod over the needs of every other database user. Using query banding and putting throttles in place, they have been able to keep query response time to less than five seconds while not overwhelming the system. “At the end of the day, you’re going to have to find out your individual sweet spot for the workload you want to run on your system,” Barrineau told the audience.