Sonic Drive-in CIO on learning to embrace big data and cloud computing

Mark another in the loss column for enterprise data warehouses. Sonic, the popular drive-in fast food restaurant, has ditched its big, heavy-duty data warehouse for — gasp! — a cloud service, according to Sonic SVP and CIO Craig Miller. And it’s opening the company’s eyes — or at least its arms — to all sorts of new possibilities.
Anyone familiar with the history of data warehouses, or EDWs for short, probably has heard part of this story before. The promise is a centralized system that stores the company’s most-important data, the stuff used for producing annual reports and tracking the metrics deemed critical to the business. But the cost and complexity of building a data warehouse means things rarely go as planned.
It’s hard to connect the various systems that feed into and from the data warehouse. It’s expensive to scale them as data volumes grow. Business intelligence software to sit on top of them and actually generate reports costs money, too. And god forbid you need to change a schema or add a new data source.
In many ways, Sonic is a perfect example of how the good ideas behind an EDW deployment can go awry.

Tracking 3,500 stores with your hands tied

One major problem is that Sonic has more than 3,500 stores across the United States and 90 percent of them are franchised. When the corporation doesn’t own the stores, Miller explained, collecting the data it needs to make company-wide decisions becomes a very big challenge. Assuming the data arrives, the next challenge is disseminating from a big system in the corporate data center to various divisions, regional offices and employees in the field that need it.
Oh, and one more thing about EDWs: They tend to be a bit inflexible. If someone at Sonic wanted to look at some different data in order to get some new perspective or see a different aspect of the business, it could takes months — up to a year — to develop and distribute that report, Miller explained. That’s because it requires a lot of programming at every level — from the BI models, down to the database schema and even the tool that pulls data from the individual franchises.
The whole process could also cost hundreds of thousands of dollars, Miller said.

A classic data warehouse architecture. Source: Wikipedia Commons

A classic data warehouse architecture. Source: Wikipedia Commons

Sonic drives into the cloud

So about two years ago, Sonic began looking for a solution to its EDW problem. It wanted to simplify data collection and storage, and it wanted to analyze new ideas and new data sources as they popped up. Some folks choose Hadoop to help address issues around schema and volume at the storage level, and most data warehouse vendors even connect to Hadoop. But it’s not an ideal solution and it’s still quite complicated.
Miller’s team decided to take a very different tack and opted for 1010data, a cloud-based analytic database, as the foundation of its new EDW strategy.
Asked about security, the elephant in the room whenever a large enterprise considers moving to the cloud, Miller said, “I wouldn’t call it concern. I would call it consideration.” Sonic did the necessary due diligence around things like security, provider accountability and lock-in, and it decided the cloud was the right move.
“Our core competency is preparing and selling and serving food,” he added. (And, I’m told by co-workers, its hot dogs are delicious; I’ve only tried the root beer floats.) If Sonic is spending more money building tools to analyze the business instead of running the business, and more time modeling and cleaning data than analyzing it, Miller added, “Then there’s a problem.”
Sonic began building its 1010data-based platform in late 2012, and the results have been impressive. Its massively parallel processing engine provides fast queries, and the system requires much less cleaning of data before uploading. Sonic is already adding in new data sources about stores, markets and weather, Miller said, and is even bringing in data from syndicated sources.
Thanks to a then-budding partnership between 1010data and Tableau, Miller said, Sonic was able to build a connection between its data store and Tableau so analysts can analyze and visualize data on whatever devices they’re toting. The 1010data and Tableau partnership will be formally announced on Wednesday.

Growing the business by growing the data warehouse

Right now, Sonic is using its new data and new platform in much the same way as it used to — measuring the success of marketing campaigns, and analyzing franchise data at the corporate level and using it to guide franchisees on how to do better. However, Miller said, the new flexibility does allow Sonic to answer specific questions that franchisees might ask of it, and the company is actually considering giving franchisees access to its data in the next year or two.
But the move to the cloud is already paying off. For example, Sonic’s operational scorecard, which analyzes data on more than 150 regional markets, now runs quarterly instead of annually, letting the company react faster to changing consumer behavior.
“We’ve saved in the order of millions and millions in many ways,” Miller said.
And, he added, “This is just scratching the surface” of what Sonic plans to do on the data front. Sonic is looking at bringing data from all sorts of sources (mobile, social media even commodity markets among them) and wants to delve deeper into predictive analytics.
“Once you have all the data and information,” Miller said, “you move from ‘what happened yesterday or last week’ … to ‘what do you think is going to happen tomorrow?'”