While many of the FCC’s broadband workshops have dealt with current, concrete issues such as the deployment, adoption, and utilization of broadband in the United States, Thursday’s FCC workshop took a refreshing departure from the here and now — which in government practices is the equivalent to three years ago — and spent time discussing the ideas that could potentially change what we know as the Internet.
One of the questions in the discussion was, “What might the Internet architecture look like in ten to twenty years, beyond incremental changes like speed increases?”
FCC Chief Technologist Jon Peha moderated the talks, and raised the specific question under this heading: “Is it possible to have multiple ‘Internets’ running simultaneously using different protocols and maybe even serving different purposes; and if so, is this a new product line for service providers?”
“Whether or not it comes to pass, it is clearly a possibility,” David D. Clark, Professor and Senior Research Scientist at the MIT Computer Science and Artificial Intelligence lab replied, “There are people in the research community who deeply believe in it. They think that the ability to take the physical resources, the routers, the circuits, etc., and virtualize them in the same way that we virtualize a machine so that you can then run different…what today we would call ‘internets,’ — different architectures — on different slices is the way to preserve flexibility in the future.
“There are technical issues there, there are also investment issues there,” Clark continued. “If I’m a facilities owner, what is my motivation to build a system like that in which I have, in fact, reduced the part of the value chain over which i have any control? For example, if I don’t control routing, how do I know where to put the physical circuits so that the logical network actually has circuits going where they want?
“There are lots of problems in that space, but the excitement of the enthusiasts there is that if it turns out that you want different internets in different places on different architectures, you can do so without having to go back and replace the equipment; and in the virtualization of the machines in the machine room, you can incredibly change the facilities upon which people can offer their services. That is a distinct possibility, it’s got problems, and I’m really enthusiastic about people that are pushing it… It might change the whole sense of what it means to own facilities, what it means for there to be a network, and I think that is a possibility, and it’s a radical one.”
Dr. Taieb Znati, Division Director, National Science Foundation, agreed with Clark and took the multiple internets notion one step further, saying multiple networks is actually the only way it could go, citing attempts at converged networks, such as ATM, that have failed.
“I think virtualization will afford people the flexibility to deploy different networks for specific purposes, some of them will be short in duration, some will be long duration,” Znati continued. “The applications will determine how the network will have to be configured in order for application to reap as much benefit as possible. Now it’s not going to be easy to do, and I think virtualization has its problems right now, like David said, and that will be the challenge for us.”