Foster Hinshaw has a theory. The most successful consumer-oriented companies—the Wal-Marts, Amazons, L.L. Beans, and Staples of the world—are successful not just because they understand their customers, but because they can operationalize that understanding. They collect massive amounts of information about past transactions and store it in data warehouses, and they actively mine those warehouses using business intelligence software.
Hinshaw illustrates with a story. “I called up L.L. Bean an hour before the cutoff time for FedEx Christmas delivery. I said, ‘I want this green sweater for my wife.’ This great customer rep says, ‘I’m sorry, that sweater is not in stock in green.’ So I was ready to run out to the mall. But then she said, ‘However, we see that your wife also likes teal, and we do have that sweater in stock in teal.’ Now, I won’t admit it in public, but I didn’t know that my wife liked teal. But somehow L.L. Bean did know that, and once she said it, it was obvious. So she was able to help me with the stuff I needed. That is what you call operational data warehousing and business intelligence.”
Hinshaw pays attention to such episodes because data warehousing has been his life for much of the last decade. In 2000 he founded Marlborough, MA-based Netezza to build data-warehousing appliances that speed up certain kinds of business-intelligence (BI) queries. Netezza’s devices are best for “deep analytics,” Hinshaw says—the kinds of questions that only a handful of PhD statisticians at each big company would even know how to ask. “They represent about 5 percent of the usage pattern in the large data market,” he says.
“But while I was at Netezza, people started asking, what about the mainstream, the other 95 percent, the guys doing routine BI?” Even before Hinshaw left Netezza (which eventually went public, in one of the top-grossing IPOs in Massachusetts in 2007), he started thinking about a new kind of appliance that would be optimized for more mainstream BI queries—questions that don’t require heavy modeling, but do require culling through terabytes of data.
And the result was Dataupia (pronounced day-TOE-pia). The Cambridge company, founded in 2005 and funded by Polaris Venture Partners, Valhalla Partners, and Fairhaven Capital, makes massively parallel data warehousing appliances that combine servers and storage with software designed to speed up BI-type queries. I got an introduction to Dataupia’s technology and its business last September during a visit with Hinshaw, and received an update a few weeks ago from Tony Sirianni, who replaced Hinshaw as CEO in early March. (Hinshaw remains active in the company as chairman of its board and its “main technical visionary,” in Sirianni’s words.)
Hinshaw’s key vision was that BI questions could be answered faster if
