Hortonworks CEO Rob Bearden opened the Hadoop Summit 2012 show his company is sponsoring with a message of cooperation rather than competition and said his company’s goal remains developing Apache Hadoop into a enterprise-ready technology platform.

Asked by John Furrier in an interview on the SiliconAngle Cube at the start of the show about the differences between the Summit and O’Reilly Media’s HadoopWorld, for instance, he said the goals of the two conferences are very similar. The main difference is that in an industry that changes radically every six weeks a lot has happened in the six months since HadoopWorld.

“It was important to us that the Summit not be just a HortonWorks infomercial,” he said. “So we gathered a group of people with different views of big data from different levels in the stack to write our goals and agenda to create a real exchange of information.”

And, he said, HortonWorks has not changed its own business plan. “Our strategy is to take totally Open Source Apache Hadoop and productize it and manage it and make it enterprise-ready at scale. It is important that we be very open in how we do that.”

Only by creating a single, open source technology can the Hadoop community achieve a critical mass of adoption of Hadoop, he said. While he mentioned no names, he said that one issue in the community is the appearance of multiple versions of Hadoop, some including proprietary elements. This can cause market fragmentation resulting in confusion and dissipation of the momentum that Hadoop needs to make the market.

The analogy can be made to the history of Unix in the late 1980s and 1990s. While the Unix core was an open standard, that core was never fully stable for production environments. The large systems vendors of the era – Digital Equipment Corp., Data General, HP, IBM – each created its own semi-proprietary version of Unix to run on its systems with the result that the Unix market was fragmented, and today Unix has almost disappeared as an operating system.

Bearden, however, focused on the positive, noting that the various startups in the Hadoop community are all contributing to the development of Apache Hadoop.

When asked about the supposed competition between HortonWorks and CloudEra played up in the press earlier this year, Bearden implied that there never was any competition. “[CloudEra CEO] Mike [Olson] and I are friends. We have lunch together. We work with CloudEra. Both companies are committed to making Apache Hadoop the next generation platform.”

Nor does Bearden see Hadoop as competition for traditional RDBMS systems like Oracle or Teradata, and HortonWorks is working with some of those vendors as well. He admits that not all the RDBMS vendors understand that Hadoop is additive rather than a potential replacement for their products but said Teradata in particular “gets it” and is working with HortonWorks to develop interfaces that let Teradata analytics structure and work with the new data sets in Hadoop.

“We need clear interfaces,” he said. “It would not be smart for HortonWorks to create an analytics capability and disintermediate parts of the core community.” It also probably would not be possible, and any attempt would draw important resources away from the core thrust of making Hadoop enterprise-ready. More than enough important unstructured data exists to create a market for Hadoop.

As for how HortonWorks makes money, that strategy also has not changed, he said. HortonWorks will provide training, support, and enterprise services around Open Source Apache Hadoop.

“What’s vitally important now is to reach a critical mass of adoption of Hadoop in the market,” he said. “To do that Apache Hadoop must be made enterprise-ready, and it must have a company behind it that can support it at the same SLAs that enterprise vendors do with their products.”