The Internet has problems; that isn’t news to most users. Slow loading speeds due to heavy traffic, hit-and-miss responsiveness to mobile devices and the ever-present fear of a major security breach are just a few of the glitches that range from annoying to dangerous and make the Internet an ongoing work in progress. However, there is only one Internet, so its denizens must either overlook the flaws by contenting themselves with its benefits or do something radical, such as invent a new one.
Unbeknownst to most Internet users, University of Kentucky computer science professor James Griffioen and researchers across the country are working on the second option.
“The Internet has been around for a long time and has been incredibly successful. At the same time, we are starting to recognize that it has some weaknesses. Some of those are currently being addressed by trying to come up with the next generation Internet architecture,” he says.
Dr. Griffioen explains that when the Internet was created, no one could have foreseen the variety of ways it is used or how dependent our world has become on it for commerce, communication, entertainment, research, etc. As a result, the architecture isn’t equipped to do what its global users increasingly demand from it. A new infrastructure is needed for what Griffioen refers to as “the next generation Internet.”
Sounds easy enough, right? Simply develop a test product, experiment with it and optimize based on the findings. Dr. Griffioen says the problem with such an approach is the scale that is required.
“How do you test out a next generation Internet architecture? You have to create something that allows developers to test their ideas. Plus, to simulate reality it needs to be around the size of the current Internet and have a lot of traffic. You also need real live users who can try it out. This is where the Global Environment for Network Innovations (GENI) project office plays a role.”
The GENI project office is funded by the National Science Foundation and administers grants to university researchers who are collaborating on the Internet’s successor. A vast partnership among American universities has led to getting a test version of the next generation Internet off the ground. Dr. Griffioen says the key to the design of this testbed is something called “virtualization.”
“We are already using virtualization with virtual personal computers. Right now, you can create an imaginary PC using a software program. Inside that virtual PC, you can load an operating system and run programs inside of it. You can actually have multiple virtual PCs running on your desktop or laptop at the same time—one running Microsoft Windows, another one running Linux and yet another running something else. We are using the same idea to build the next generation Internet. That is, we can virtualize the routing infrastructure and run several virtual routers on that one physical router. By doing so, I can have my own network—which I have specifically tailored for my purposes—and someone else can have their own network right next to mine, yet we will be using the same infrastructure.”
Dr. Griffioen says such networks can be set up within minutes, which leads to questions regarding the performance quality of such networks. How can a developer know whether his or her network is effective? Dr. Griffioen, fellow computer science professor Zongming Fei and the research team, in partnership with Indiana University, have set out to answer such questions through a system they have developed called GEMINI, as well as a product called the GENI Desktop.
“The idea behind GEMINI is to see how well someone’s network is performing—how fast traffic is getting through, how much data is being sent between the nodes and does any of it actually work? GEMINI will take a network that has been built and add in the infrastructure to capture everything that needs to be used to measure a network. Finally, we package it within a helpful graphical user interface, which we call the GENI Desktop.”
When asked what has surprised him most through the research and development process, Dr. Griffioen hinted that the work on the next generation Internet is only scratching the surface and that the delineation “the Internet” might disappear.
“This ability to create virtual networks under one infrastructure is a very powerful abstraction. In other words, there may not be just one Internet; there may be multiple specialized Internets, each one run concurrently over the same infrastructure. Some might be optimized for video conferencing. Others might be ideal for distributed gaming and others uniquely suited for high-speed, secure transactions. Each type of network would have its own characteristics.”