Yahoo, which needs to turn a buck as well as focus on its own issues, is spinning out its Hadoop unit into an independent new company co-funded by Benchmark Capital called Hortonworks after the Dr. Seuss character Horton the Elephant.

It's a nice whimsical touch considering Hadoop is really a stuffed elephant who had his identity stolen, which may explain why the code is sometimes cranky.

Hortonworks will compete against other Hadoop commercializers like Cloudera, IBM, newcomer MapR and its buddy EMC and figures to accelerate Hadoop's adoption by making the code "more robust and easier to install, manage and use for enterprises and technology vendors."

It talks of Hadoop becoming the "de facto platform for managing and analyzing Big Data" and anticipates that more than half the world's data will be stored in Apache Hadoop in five years.

Since Yahoo, inspired by Google's MapReduce model, developed Hadoop, Hortonworks promises to be staffed with many of the key architects and core contributors to the open source Apache Hadoop project, "the same team," it says, "that pioneered Apache Hadoop and delivered every stable release from 0.1 to the most current release." Hortonworks can obviously out-credential its rivals.

Figure it's starting with around 25 people and means to hire.

Eric Baldeschwieler, formerly VP of software engineering for the Hadoop team at Yahoo, has been named CEO of Hortonworks. Rob Bearden, now at Benchmark but previously COO of both SpringSource and JBoss, two successful (and ultimately acquired) commercial open source companies, will be handling the business side as COO again.

Yahoo and Benchmark, VC to Red Hat, MySQL, SpringSource, Twitter and eBay, didn't disclose how much money they're stuffing in the place or who owns what.

Yahoo, which will reportedly be a "strategic partner" of Hortonworks, says it intends to continue its internal investment in the key widgetry since Hadoop is now behind every click on Yahoo, currently processing five billion jobs a month on 42,000 servers handling its content and advertising.

Hortonworks means to sell subscriptions, training, certification and support. It says it "may build applications or tools in the future that are only available to paying customers," but it's committed to all core code being open source. It says it will donate all source code back to Apache. Apparently it means to design solutions for specific industries.

It says reference implementations should come from Apache and that Hortonworks "will do its development in public, releasing early and often via Apache."

It will "continue to be a driving force behind Apache Hadoop and major contributors to core projects including Hadoop Core, MapReduce, Hadoop Distributed File System (HDFS), ZooKeeper and Pig."

Evidently it figures to make breakthroughs in performance, availability, scalability, security and integration and "work with the community to add new capabilities to open source including administration and monitoring."

Hadoop is used by companies like Facebook, eBay and Netflix.

• BigDataExpo New York Speaker Profile: Eric Baldeschwieler – Hortonworks
• Old Big Data Cozies Up with New Big Data
• Hadoop Turns Seven