In the next few years, a project known as PlanetLab — started in part by computer science professor Larry Peterson — could revolutionize the Internet.
The founders of the project, which the University now hosts, have called it one of the most "exciting" projects in the history of the Internet.
PlanetLab comprises a patchwork of computers around the globe which forms a private network through the Internet. Its purpose is to provide a testbed platform for developing new and improved Internet services and networking capabilities.
Other similar research projects have remained closed systems because they do not make use of the Internet. PlanetLab's value, however, comes from how it expands existing Internet capacities to meet the needs of researchers.
Its goal is to allow both academic and private sector organizations to develop a secondary channel of interaction without bringing down the current system.
Peterson compared PlanetLab to the vision researchers had during the early days of the Internet — a network free of commercialization.
"The Internet is now so commercial and it's difficult for researchers to try out new ideas," he said. "It was very different in the early days of the Internet. Basically, we're trying to recreate the early days of the Internet with PlanetLab."
The type of service PlanetLab offers is classified in the industry as an "overlay" network. These networks add capabilities to the Internet without changing the Internet's existing structure.
"[PlanetLab] is not a replacement," added Peterson. "That's not the right way to think about it. It is more of an incubator for new ideas. That [characteristic] will give the project more functionality."
Problems are appealing
Part of the appeal of PlanetLab is that the network experiences all the behaviors and problems of the real Internet — including lag, viruses, and server crashes. These complications force researchers to work in a realistic environment.
The diversity of the network is also appealing to researchers. The PlanetLab network extends across different technical properties, network capabilities and geographical boundaries. Thus, Internet solutions developed through PlanetLab will have the luxury of a wide array of test environments.
Researchers involved in the project are equally impressed with the rapid success of PlanetLab.

"The fact that so many people are involved helps make [PlanetLab] a very exciting project overall," said Scott Karlin, a research associate in the University's computer science department who is currently working on the problems of coordinating computers on the network.
Karlin's work is one of close to a dozen projects on Princeton's campus making progress in Internet research through PlanetLab.
Researchers are trying to improve content distribution networks and distributed query processing engines.
Fundamentally, though, PlanetLab relies on a technique transparent to users called "distributed virtualization." This feature allows each user to operate an application that is actually distributed throughout the entire worldwide network. Peterson explained distributed virtualization as a user interface illusion.
"When you use it, you think you have it all to yourself," Peterson said. "The machine has been virtualized to make you think you are the only one with access to it, but the application could be running simultaneously on other computers around the world."
'Diverse coverage'
To host a computer on the PlanetLab network, research institutions need to fill out a simple online request form on which they indicate their research interests.
There are about 200 nodes worldwide in about 100 locations. The majority of the sites are located in the United States, Europe and Australia.
Peterson said PlanetLab has a goal of attracting 1,000 nodes within two or three years.
"The target was based on a couple things. We researched organizations worldwide that might be interested in hosting the system and figured a certain fraction of them would join.
"Diverse coverage of the Internet is one of the primary goals of PlanetLab, and we feel that 1,000 would be sufficiently broad in two years."
The software run on the nodes is a modified version of the Red Hat Linux operating system. The Linux operating system was chosen by PlanetLab's architects for the system's ability to process large of amounts of information and its quick setup time.
PlanetLab has a relatively brief history. The majority of progress on the project occurred within the past year among a core group of researchers.
"PlanetLab began as a collaborative effort," Peterson said. "Myself and a colleague at Intel concocted the idea and took the initial work to a group of academics. It was really a grassroots beginning."
In 2002, Peterson and University of California at Berkeley professor and Intel researcher David Culler invited roughly 30 leading network and distributed systems researchers to help them lay the theoretical foundations for what grew into PlanetLab. The group was charged with developing an international computer network to expand research opportunities on the Internet.
"It was immediately apparent to the people in the group that [the idea] was a good thing," Peterson said.
As soon as the group worked out many of the project's details Intel stepped in to provide the seed funding to get the project off the ground. From the very beginning, Intel was very supportive of the idea, Peterson said.
The Intel funding opened the door for expansion of the network by adding computers needed to make PlanetLab a successful experiment.
"It is a platform for researchers to supplant what you think of as the Internet today," Peterson said. However, Peterson notes that PlanetLab can extend to nonacademic arenas. If the idea is successful in the long run, it may become the basis of similar networking experiments in the commercial sector.
"We think these kinds of services are going to become part of the global communications infrastructure," Peterson said. "We want to make sure we build a process so that the research community never gets shut out again."
Researchers at Princeton hope that the energy surrounding the project sustains itself.
"Generally, progress is good," professor Vivek Pai said. "We get about a million accesses a day. And that's without a serious amount of advertising."
The early successes of the program are uplifting to those involved.
"Part of the reason this has taken off so well is that this is something that is needed," said Peterson.
In the upcoming months, Peterson hopes to have Princeton host a consortium that would discuss industrial involvement in the project. The group that convenes would create a plan of action to invite other companies and educational institutions to participate in the project.
Already, Intel and Hewlett Packard have earmarked money to set up nodes at various research and development centers around the world. Peterson said he believes that PlanetLab's future now depends on similar private high-tech companies' involvement to fuel the expansion.
The real world impacts of PlanetLab can already be felt on Princeton's campus. Peterson plans to use PlanetLab in his graduate seminar COS 597B: Advanced Topics in Computer Science: Planetary-Scale Services. He intends to invite select undergraduates to take the course.
In the months ahead, as PlanetLab expands its operations, consumers may see the effects of the innovative research.