posted by Jordan Spencer Cunningham on Fri 2nd Oct 2009 19:50 UTC
IconA while back, we covered the release of the free Cloudera distribution of Hadoop-- handy software to manage data across a multiplicity of servers-- the same software behind Yahoo!, Facebook, and other successful companies. Though Hadoop and Cloudera's Hadoop have been truly stellar at what they do, it's all essentially been done via command line, which for many people isn't the most productive or user-friendly type of interface. The folks at Cloudera knew this, so they've gone ahead and created a graphical interface to communicate with Hadoop.

Though experienced users of the software may prefer a CLI, Cloudera Desktop is designed to make Hadoop easier to use for new and unexperienced users as well as to make working with the software more efficient for experienced users. It's essentially a graphical interface for Hadoop that runs in a web browser on Windows, Linux, or Macintosh; because it runs in a web browser, there is no installation required on the client computer. Cloudera has posted a video and some screenshots of what the application looks like and how it works along with an introduction to the tools included with the software.

The software itself supplies the user with varied tools to run and monitor a Hadoop cluster among which are the file browser, which copies and browses files saved in a cluster; the job submission tool, which creates, runs, and saves jobs; the job browser, which keeps track of jobs and their statuses; and the cluster health dashboard, which monitors the health of a cluster and displays alerts if problems occur. These are just the initial four tools included in the first release; Cloudera plans to add more features and tools as time marches on, and updates and upgrades are all a part of the package. Jeff Hammerbacher, VP, Products at Cloudera, said that "over time, we expect that individual companies will find different collections of applications useful for their needs, and they'll have the ability to add and remove applications at their discretion."

For now the public beta only works with the Cloudera distribution of Hadoop. Hammerbacher said that they have submitted most of the needed patches to make Apache Hadoop compatible with Cloudera Desktop, but they're currently waiting for some of those patches to be approved while others have been committed. "So initially, the answer is that Cloudera Desktop only works with Cloudera's Distribution for Hadoop. In the long term, we'd like to see Desktop work with all versions of Hadoop."

Cloudera Desktop, like Cloudera's distribution of Hadoop, is completely free and can be downloaded as such. Support for Desktop and Hadoop alike can also be found at Cloudera's support page from both the official company and the community. The server-side application is available for Red Hat and Debian variants of Linux as well as Amazon's EC2. A tarball is also available for those who prefer to compile the server from the source code. Once successfully set up on the server, no more work is needed for the client computer except to log on from the web browser. For those who wish to have a feel of what Desktop is like before trying it on one's real-world server, one can download a VMware Image.

In other news, Cloudera has also been working on Squoop, which is a tool that imports other types of databases into Hadoop. Hammerbacher also said that the Cloudera team would like to develop applications for Hive and Pig as well. Also, those interested in window managers can actually get the BSD-licensed code at the core of Cloudera Desktop's window manager to play with.

e p (2)    No Comments

Technology White Papers

See More