InformationWeek: How are you using HP Neoview?
Stewart: For our Retail Link. We use this system to start answering Retail Link questions from suppliers. Those folks want to see sales performance of items.
InformationWeek: So Neoview is in production?
Stewart: It's absolutely in production. It is meeting our performance and availability requirements. Wal-Mart has very stringent requirements in that area. We would not have participated in this discussion without having that system performing within Wal-Mart's data environment.
Wal-Mart has such an extensive requirement for use and understanding that the Neoview system fits very nicely into that configuration along with Teradata and others.
InformationWeek: Is Neoview replacing any part of Teradata?
Stewart: It compliments our current environment as we continue to grow and innovate in this space. Wal-Mart is a 7-by-24-by-365, better-than-Six-Sigma shop. Neoview fits right into that environment of extremely high availability and high performance. It performs extremely well. It has management characteristics where we can manage the plex very well.
We are a low-resource shop. We basically always need extremely high performance of high levels of systems management for our environment. Our employee-per-operation MIP is basically not on the chart relative to how many folks we have. We have a very small team of operations to support the plex.
InformationWeek: Are you expanding your use of business intelligence at Wal-Mart?
Stewart: There's a whole new set of queries and algorithms we're looking at and will continue to look at as we expand our inventory of formats. As we expand our product line. As we look at product line for best fit for store-of-community premise. How do we best model for store-of-future, store-of-community, and ensure we have the best layout to support that function? As you can imagine that's a ton of crunching.
InformationWeek: So business intelligence is big for Wal-Mart?
Stewart: Business intelligence is huge. It is huge. My colleague Marc Rosen has established a function under a direct report to lead that area right now.
It's really the fact of how do you harness all that data and have the right set of algorithms to do the level of interpretation to make the best business decision. We sit on six, seven petabytes of data and look at that across our spectrum. [There is] one table that has four hundred billion rows. What are you going to do with that table in order to make a business decision?
We collect data on everything. Eight hundred million transactions and 30 million people a day. We send all relevant information back to stores at the end of day, so you're seeing how much data gets pumped out each day and now you're going to get a BI framework over that. Really, how is it you're going to harness the data to drive the right information so [you can] use that info to have some sort of framework to get the right decision against it?
Were making significant investments in BI tools. Jim [Scantlin, director, enterprise information management] is ensuring we have back-end capability so we can do the tiered approach. We're strongly considering data marts and looking at a master data-management focus. There are quite a few initiatives.
InformationWeek: Are more people using BI internally?
Scantlin: A lot of people in the company are asking for quicker and easier access to data. We want to make sure it's readable and usable by internal customers.
InformationWeek: So are making available BI dashboards, things like that?
Stewart: Dashboards, visual graphics, rendering tools to make a business decision. One of our examples is a store layout and how we get the best product value as we put product out. Get a what-if and decide what you want to do, then have your store planning focused.