Hadoop vendor MapR has raised $110 million in capital to help it keep growing in a heated enterprise Hadoop market. Most of the funding — $80 million — came in the form of venture capital from Google Capital(s goog), Qualcomm(s qcom) Ventures and MapR’s previous investors Lightspeed Venture Partners, Mayfield Fund, NEA and Redpoint Ventures. The remaining $30 million was debt financing.
The cash adds another angle to the question of which companies will end up dominating the Hadoop space. It comes just a few months after Cloudera raised more than half a billion dollars and a strategic partnership with Intel(s intc), and after Hortonworks raised a $100 million round. MapR has now raised $139 million in equity capital since it was founded in 2009, and CEO John Schroeder (pictured above) said it’s prepping for an initial public offering as are its more well-known competitors. He claims the company, which is based more on license revenue than on support, will be profitable in the next several quarters.
MapR has been on the receiving end of potshots from competitors for years because it’s based on proprietary file system and management technologies, but Schroeder said that concern isn’t having much of an effect on customers. “They can be fooled by all sorts of vendor Olympics like that” before they’re spending real money on production Hadoop systems, he said, but customers that are dropping lots of cash have done their research. MapR is about 80 percent open source while Cloudera is about 85 percent open source, he estimated, and buyers see through the argument of “do open source, but just buy it from us.” (The Hortonworks distribution is entirely open source.)
Schroeder thinks MapR still has a technological edge with its proprietary tools and he says it also offers customers the most choice in the components they want to use. MapR, for example, supports a handful of SQL-on-Hadoop tools including Hive and Drill — a project it is spearheading — but also the Cloudera-developed Impala and HP’s Vertica software. Like its competitors, MapR also supports Apache Spark, the increasingly popular processing framework that promises a faster, easier way than MapReduce to write big data jobs. (Although he says it’s “way too early to draw [the] conclusion” that Spark will overtake MapReduce in importance.)
MapR does anything that extends the application side in open source, while trying to create stickiness with better software around performance, management and operations, Schroeder claimed.
MapR chose to accept capital from Google and Qualcomm from a long list of investors “looking to write $50 to $100 million checks” because they represent the future of where big data is heading, he said. Google was responsible for inventing many of the technologies on which various Hadoop components are now based, and MapR has a close relationship with Google as a partner for running Hadoop on its Compute Engine cloud. Curiously, Google is trying to move away from its MapReduce roots with its new cloud Dataflow service, but Schroeder doesn’t see that having too big an effect on the Hadoop software market in the near term.
Qualcomm, he said, has unique insights into the internet of things, which is a space that will drive companies to store and analyze a lot of data.
Handicapping the Hadoop market just keeps getting more difficult; with so much capital flowing into it, the companies fighting for supremacy seem to have little difficulty finding enough to keep running. And despite speculation over which large-cap software vendors tried to acquire which Hadoop vendors, and criticism lobbed at one another over technologies or business models, no company in this space is backing down. They all make valid claims, and they’re all talking about becoming public companies.
Schroeder makes a fair comparison to the relational database market decades ago and the bloodsport that was doing business in it. He predicts two Hadoop companies will make a bunch of money “and one will still being hanging around 20 years later.” Guess which one he thinks that will be.