• United States

GENI looks to conjure up next-generation network

Mar 20, 20063 mins

The Global Environment for Networking Innovations is still in the planning stages, but the next-generation research facility is taking significant steps toward becoming a reality.

About 120 people from universities and commercial labs around the country recently gathered at a town hall meeting in Arlington, Va., to discuss the conceptual design of GENI, why it’s needed and what type of research might be conducted over the network.

GENI will be a National Science Foundation (NSF) experimental facility that will let scientists conduct research that goes beyond the constraints of today’s Internet, says Larry Peterson, chair of the GENI Planning Group and professor and chair of the computer science department at Princeton University.

NSF first proposed GENI last summer. GENI researchers will not be looking simply for “a new version of IP,” Peterson says.

A facility like GENI is needed for many reasons, he says. One is that the industry will not solve the problems of today’s Internet because “there’s not incentive” to do so. Peterson says another reason is that the academic community views any research as “risky” that’s not backward-compatible to today’s Internet.

The original GENI design includes a national and eventually international fiber-optic network with programmable routers, clusters at the edge sites, wireless subnets and peering to the Internet at MAE-East and MAE-West. Peterson says this will be necessary for researchers to have access to the vast amount of content on today’s Internet.

GENI is still conceptual, however. It is expected to take five to seven years to build, and a construction date has not been set, primarily because that date is tied to funding that needs Congressional approval and is still one to two years off.

Even though the build will take several years to complete, Peterson points out research could get started within the first year of construction. “We continue to prototype various technologies that GENI will leverage. We expect these prototypes to be sufficiently advanced to give users some of the capabilities very early in the construction phase,” he says.

Peterson also talked about some of the requirements of GENI, which include architectural and service neutrality, virtualization and real users.

The group also is putting together the GENI Community Consortium (GCC), which will run like the IETF with its working groups. The GCC’s working groups, which are focusing primarily on design at this point, are in the areas of research coordination, facilities architecture, backbone network, wireless subnet, distributed services and education and outreach.

One of the prime goals of GENI is to “change the nature of networked and distributed systems design,” Peterson says.

He says that doesn’t mean the current Internet gets tossed out; instead, the group hopes to design a future Internet that’s more secure, available, manageable and better suited for computing in the next decade.

Other well-known experimental networks such as Internet2 are expected to continue to coexist with GENI.

“Internet2 provides great value to the broader scientific community,” Peterson says. “It has been used to deploy new services, but it has not met the needs of the networking research community. This is because it supports real users, and so cannot tolerate the disruptive research we want to do. GENI explicitly tries to break this dilemma by supporting both clean-slate design and real users.”