Is Cloudera the Next Oracle? CEO Mike Olson Hopes So
Cloud computing is expected to commoditize entire sectors of the IT industry over the next few years, and to create entirely new classes of big companies as well.
One of those big companies will be Cloudera, says Mike Olson, the firm’s CEO. Or at least he hopes it will be Cloudera, a Palo Alto, CA-based startup that Olson, a database expert, is modeling after Oracle. The two-year-old startup raised $25 million last week in a Series C round led by Meritech Capital Partners.
“For 22 years, I watched Oracle,” Olson said in an interview this week. “I competed against Larry Ellison for a bunch of years before I finally sold him a company [Sleepycat Software, an open source embedded database engine] in 2006. What Oracle did in the ‘80s and what it’s done since—that lesson is not lost on me. I think Cloudera can be every bit as transformative.”
Cloudera sells tools and consulting services so that what Olson calls “ordinary IT staffs”––people who are trained in computing but don’t have PhDs in computer science-–-can run Hadoop, an open source implementation of a platform that Google built around 2004 to store and analyze the enormous amounts of data it was generating from its search indexes and logs.
Although he didn’t realize it at first, Google was solving a problem that had never been solved before, Olson says, because the problem hadn’t existed: large-scale analysis of data, which Google engineers figured out how to distribute over a large network of standard computers instead of loading it all onto one big machine.
Google’s system didn’t interest him much at first––it was missing a lot of the transaction processing features of high-end relational databases, he thought, and most of the database industry ignored it. But after he’d worked at Oracle for a couple of years, Olson says he realized that relational databases-–-which link data by common characteristics, breaking the hierarchical tree structure for data used by mainframes—weren’t built to solve the problems that Google was trying to solve.
Furthermore, he says, he saw that “Web properties weren’t any different from ordinary enterprises. We were five years early, but everybody––banks, insurance companies––would have this same data problem. Data was so big and so common, and it was growing so quickly, that these techniques would need to move into the enterprise.”
Olson co-founded Cloudera two years ago, in October 2008, along with Jeff Hammerbacher, who’d worked with Hadoop at Facebook, and Amr Awadallah, who’d worked with the software at Yahoo. A fourth co-founder, Christophe Bisciglia from Google, has left Cloudera, while Doug Cutting, who led the team that built Hadoop at Yahoo, joined last year.
Accel Partners invested $5 million in Cloudera when it launched––just as the financial markets were tanking—and the company raised another $6 million from Greylock in 2009, and the $25 million last week. Cloudera now has several large customers, including eBay and JPMorgan Chase.
“We don’t really need the money,” Olson says, “but there are tactical and strategic reasons to raise it, and strong interest from investors…Now I can afford to hire before I need people, and if I look around and see interesting technology, I can make small strategic acquisitions. There’s nothing planned, but it’s added options.”
But is Oracle the company for Cloudera to emulate? Even Oracle––along with Cisco, HP, Dell, and EMC––is vulnerable to the sweeping changes that cloud computing will create over the next decade, said Mark Leslie, a lecturer in management at Stanford Business School and the founding chairman and CEO of Veritas. Leslie was speaking Wednesday at the VentureWire FASTech conference in Redwood City, CA.
All these companies “are competing to build a vertical stack, and I think the cloud profoundly disrupts that,” Leslie said. “Ten years ago the pre-eminent server was the Sun Starfire 10000, and it’s gone. Everybody buys dense, small packed chips, and I think the same thing happens in networking, storage, and security.” And, Leslie added, these new solutions may come from entirely new vendors.
But Olson says he’s worked with open source software for two decades, and Hadoop to him is a milestone. It’s the first time that open source software has been used to create something new instead of just commoditizing an existing market, making “an expensive, well engineered knock-off and [giving] it away,” he says.
Other new companies like Karmasphere and Datameer are also springing up to fill holes in Hadoop, and Cloudera is forming partnerships with them along with more established players like Informatica. “We believe this platform will be as successful as the relational database was in the early ‘80s, back in the day when radical new technology was needed,” Olson says. “Oracle and IBM and companies like Ingres all invested energy in making that platform reliable and let [independent software vendors] and [original equipment manufacturers] build the apps, and that’s what we’re trying to do now.”