Google, IBM Plant Distributed Computing Seeds

Partners IBM and Google will allot servers and open-source software to help university students learn parallel computing.

Computer science students are getting a chance to learn about distributed computing from two of the biggest names in high tech.

Google and IBM said they will provide servers and open-source software to help students and researchers at colleges and universities write software to build out the Internet.

The ACCI (Academic Cluster Computing Initiative), announced Oct. 8, aims to improve computer science students knowledge of parallel computing methodologies to plant the seeds for large-scale distributed computing.

The important thing is "no longer making processors faster, its making them more parallel," Christophe Bisciglia, senior software engineer at Google, based in Mountain View, Calif., and co-founder of ACCI, told eWEEK. "Youre seeing multicore technology; networks are getting cheaper and faster so youre connecting more computers in parallel. You dont scale your software with a faster computer, you scale it by adding more computers, scaling horizontally."

Bisciglia said even the most talented students dont get exposure to large-scale computing as undergraduates. After all, this type of computing is horizontal and often involves hundreds or thousands of small servers, which universities cant afford or make room for.

Operating in parallel, such machines can enable software, such as Internet search, online social networking tools and mobile e-commerce, to function well in an on-demand Web environment.

IBM and Google both see the Internet as an infrastructure platform, or a cloud, on which to build sophisticated Web applications.


Read more here about companies competing in the area of computing in "the cloud."

IBM currently offers Lotus Connections, a social computing and collaboration software suite, while Google is attacking the market with Google Apps. Microsoft,, and others are working on similar projects.

UW (the University of Washington), Carnegie-Mellon University, the Massachusetts Institute of Technology, Stanford University, the University of California at Berkeley and the University of Maryland are piloting the ACCI.

The program has its roots at UW. Bisciglia, a former UW student, hatched the program with Ed Lazowska, Bill & Melinda Gates Chair of Computer Science & Engineering at the University of Washington, to better prepare future computer science programmers for Google-class computing.

Bisciglia got approval to bring a cluster to UW and work with faculty to create a course, "Problem Solving on Large-Scale Clusters." Once students got word that Google was involved, registration for the class filled up in 4 minutes, he said.


Click here to read more about Amazons Web services.

IBM and Google said they have allotted a cluster of several hundred computers, including Google systems and IBM BladeCenter and System x servers, that will scale to more than 1,600 processors.

The machines are in a data center at an undisclosed location, Bisciglia said, noting that Google reserves the right to move the machines around. Students access the cluster via the Internet through a URL taking users to the host name of the head node.

The servers will run the Linux operating system, Xen systems virtualization, Apaches Hadoop project, an open-source implementation of Googles MapReduce and the Google File System.

The ACCI also includes a university curriculum developed by Google and UW focusing on parallel computing techniques under a Creative Commons license, along with an open-source software add-on designed by IBM to help students develop programs for clusters running Hadoop.


Check out eWEEK.coms for the latest news, views and analysis on servers, switches and networking protocols for the enterprise and small businesses.