An Inside Look at Whole Foods Move to the Cloud
In a session yesterday at the Teradata Partners Conference in Anaheim, Calif., Ken Casey, data warehouse coordinator, and Caden Schaefer, senior developer/database analyst, both of Whole Foods Market, detailed the grocery retailer’s database move to Teradata IntelliCloud, the analytics solutions company’s secure managed cloud offering that provides data and analytic software as a service (SaaS).
Whole Foods was seeking a data solution for its sales, product and customer data that was able to solve batch load timing; solve IO challenges; solve pending capacity challenges; and shorten application run times. In short, Whole Foods’ decision makers wanted access to data, primarily sales, more quickly.
“There was tension between business users and the data team on downtime on sales data,” Casey said. “What do we do with a system that’s aging out?”
Speed was the impetus for Whole Foods decision to move its data warehouse to the cloud, along with increased capacity. The retailer has seen exceptional gains in speed from its move to the cloud. For example, the speed to perform the biggest run jobs decreased from 30 hours to two hours. Furthermore, with its previous data warehouse system, Whole Foods was seeing time outs of the system; it hasn’t seen one since it shifted to Teradata’s IntelliCloud.
In addition to the efficiency gains from its move to the cloud, Whole Foods has seen cost savings from not having to maintain capex hardware systems and datacenter operations.
Tips for Moving Your Data System
Casey and Schaefer explained the transition process from Whole Foods’ data servers to IntelliCloud, and offered tips to attendees if they should embark on a similar project:
- Work early with network/circuit providers to schedule accordingly.
- Review the application stack and work with all your vendors, application administrators, and the cloud provider’s engineering teams to fine-tune the application optimization.
- Review data volumes and plan scenarios by knowing your options — e.g., cold data vs. hot data.
- If you’re moving to a bigger system (as is overwhelmingly the case), increase spool size.
- Fallback is required; watch your space.
- Have a backup schedule, coordinating this with cloud operations early in the process.
- Rehearse the cutover Hollywood-style. Much like a group of actors sitting around a table reading their scripts, the Whole Foods team in charge of the data cutover got together in a room and read their list of responsibilities to each other before the date of the cutover. This was an important step, as we caught some last-minute changes during this exercise, Casey said.