An open-source storage engine called Kudu could soon be on the way from Cloudera, offering a new alternative for companies with big data stores to manage.
Kudu will be offered as an alternative to the popular Hadoop Distributed File System and the Hadoop-oriented HBase NoSQL database, according to a VentureBeat report, which cited a slide deck on Kudu’s design goals.
A small Cloudera team has reportedly been working on Kudu for the past two years. The company has already been pitching it to customers and plans to release it as Apache-licensed open-source software at the end of this month, VentureBeat said, citing a source familiar with the matter.
Cloudera declined to comment for this story.
Cloudera offers an enterprise data hub built on Apache Hadoop for managing big data. Now, it apparently hopes to disrupt the traditional data warehouse market with open-source software that's designed to enable commodity storage to support high-performance applications for purposes such as real-time reporting, time-series data reporting and model building, noted Charles King, principal analyst with Pund-IT.
Kudu could be a logical next step in Cloudera's evolution, said Rob Enderle, president and principal analyst with Enderle Group.
"Cloudera is a fascinating company," Enderle said. "Founded by technical experts out of Facebook, Google and Yahoo, headed by near-legend Tom Reilly and funded by Intel, this firm pretty much owns the Hadoop market at the moment."
Since Hadoop is a high-performance, high-availability storage framework, it would be a natural progression for the company to move toward a complete cloud solution with Kudu, Enderle said.
"This would be a huge step toward taking the company from a point technology firm to a full solution company and critical to its growth and long-term success," he added.
It's a rapidly evolving area, particularly given the growing need to accommodate streaming data, noted Carl Olofson, a research vice president with IDC.
That demand can create higher performance requirements than Hadoop can always keep up with using the conventional Hadoop Distributed File System, Olofson noted.
"We can expect a continuous stream of changes in this space," he said. "There's a lot going on."