IBM, Google, and colleges talk cloud projects

Sponsored by IBM, Google, and the National Science Foundation (NSF), university researchers are gathering Monday to tout their respective cloud computing projects as part of the CLuE (Cluster Exploratory) program.

In presentations at the Computer History Museum in Mountain View, academics are hailing projects like "Scaling the Sky with MapReduce/Hadoop" at the University of Washington, "Commodity Computing in Genomics Research" at the University of Maryland, and "Dynamic Provisioning of Data Intensive Applications" at the University of California, San Diego. The meeting is for CluE Principal Investigators. IBM and Google presentations also are featured at Monday's event.

[ Find out what's different about building cloud apps. ]

Funding for the efforts is being provided by the NSF, while Google is providing access to a 2,000-computer cloud and IBM is offering software, such as its Tivoli management software and WebSphere middleware.

"I really believe that [cloud computing is] the tool that will help us launch the era of modern computer science," said IBM's James Spohrer, director of Global University Programs at the IBM Almaden Research Center in San Jose, Calif.  This era involves massive data sets and processing and analyzing them, he explained.

IBM's own Smarter Planet initiative is about increasing the number of sensors in the world to get more data about such subjects as transportation systems, water management, and waste management. With streams of data coming in, only powerful new tools like cloud computing will enable processing of the data and getting useful results, Spohrer said.

IBM and Google had announced in October 2007 a pilot phase of "Academic Cloud Computing Initiative," which offered access to computer cluster running the Hadoop open source distributed computing platform inspired by the Google file system and the MapReduce programming model. In February of last year, ACCI partnered with NFS to grant funding to academic researchers to explore large data applications, resulting in the CLuE program.

In an interview, IBM and Google officials said cloud computing differs from previous concepts such as grid computing and utility computing in that the cloud relies on Internet access, offers high bandwidth and can analyze massive data sets.

Among other presentations scheduled  Monday include: "Relaxed Synchronization and Eager Scheduling in MapdReduce," by Purdue University; "Large-Scale data Cleaning using Hadoop" by the University of California, Irvine; "Indexing Geospatial Data with MapReduce," by Florida International University amd "Towards Interactive Visualization in the Cloud," by the University of Utah.

Original story - www.infoworld.com/node/94523