Cloudera cranks out fresh Hadoop distro

Cloudera has released a third version of its open source Hadoop distribution, boasting that the distro tightly integrates several other Apache-licensed projects designed to run in tandem with the distributed number-crunching platform.

Announced on Tuesday, the Cloudera Distribution of Apache Hadoop version 3 (CDH3) includes not only the Hadoop distributed file system (HDFS) and Hadoop MapReduce, the number-crunching platform that runs atop HDFS, but also Hive (a SQL-like query language developed at Facebook), Pig (a lower-level language developed by Yahoo!), HBase (a distributed database), Sqoop (a MySQL connector built by Cloudera), Flume (a data-loading infrastructure developed by Cloudera), Oozie (the Hadoop workflow system), Hue (a graphical user interface), and Zookeeper (a means of juggling distributed services from a central location).

"We've really evolved the distribution from being one where it packaged-up core MapReduce and HDFS to a distribution that tries to provide a complete solution for running Hadoop within an organization," Cloudera vice president of product Charles Zedlewski tells The Register. "If you look at large organizations with major Hadoop deployments, you'll find that they they use a superset of components that includes a variety of tools that lets you ingest data, author jobs, review results, etc.

"With version 3, you can get those technologies, all integrated, all combined together, and all 100 per cent open source."

In particular, Cloudera says, it has extended the Hadoop authentication and security model throughout this stack of Hadoop and Hadoop-friendly platforms.

Cloudera also claims that "small" MapReduce jobs run up to three times faster with the new distro, and that file system I/O is up to 20 per cent faster, with a 2X performance boost for HBase query input. It has added a new ODBS (open database connectivity) driver to integrate business intelligence (BI) clients such as Microstrategy. And via a new adapter framework, you can build connectors for importing from and exporting to specific relational databases.

The Silicon Valley outfit also offers a for-pay Hadoop product, which includes services and support and augments the Cloudera open source distro with proprietary management, monitoring, and administration tools.

Based on Google’s proprietary software infrastructure, Hadoop is a means of crunching epic amounts of data across a network of distributed machines. Named for a yellow stuffed elephant that belongs to the son of project founder Doug Cutting, the platform now underpins online services operated by everyone from Yahoo! to Facebook and Twitter, but Cloudera is promoting its use in the enterprise as well.

Hadoop mimics GFS, Google's distributed file system, and MapReduce, the company's distributed number-crunching platform. In 2004, Google published a pair of research papers on these infrastructure technologies, and Doug Cutting seized on these to build a platform that would back Nutch, his open source web crawler. Hadoop was open sourced at Apache, and it was bootstrapped by Yahoo!, which hired Cutting in 2006. He now works for Cloudera. ®