Back in the 1990s, universities kept the education network innovation cycle rolling quickly, but in the last decade or so, that cycle has slowed considerably as higher education has been saddled with legacy protocols and proprietary technology.
But operators of these research and education networks are trying to pick up the pace with an advanced networking technology pilot. The effort, led by Internet2, a not-for-profit organization made up of university members, aims to bring back an environment where innovation will thrive and new ideas will take root.
"The more we can move toward an open standards-based, vendor-agnostic solution, the faster you can mix and match technology, new companies can come along and create new technology, and it really will move the whole ecosystem faster," said Eric Boyd, senior director of strategic projects for Internet2.
And that's where Internet2 and five universities are headed with the pilot of a network operating system and IP peering application. The five universities and the networks they support include Duke University, Florida International University, Indiana GigaPOP supported by the Global Research Network's Operations Center at Indiana University, Mid-Atlantic Crossroads operated by the University of Maryland - College Park, and the University of Utah.
By developing these open-source, software-defined tools, the non-profit Open Networking Lab (ON.Lab) is helping to create a more programmable network that others can build on, customize and use with any brand of hardware.
"The more programmable you make the network, the more likely you are to be able to open up innovation opportunities going forward," Boyd said.
|'Cloudification' Is the Future of the Internet|
Internet2 has had a software-defined network for nearly three years, which means it separated the control plane that determines where to route information packets from the forwarding plane that sends those packets on their way. This separation allows network engineers to program the control plane and manage what it will do by using software called a controller. The controller acts like a lifeguard at a swimming pool by coordinating different activities on the network and separating them into different lanes.
In order to communicate, the control and forwarding planes use an OpenFlow interface that's vendor-neutral. Internet2's 15,000-mile network includes 38 OpenFlow switches.
Ultimately software-defined networking is an economic reality that can't be stopped, and it's prompting a fundamental shift in how networks operate.
"We've gotten to the point where we have so many mobile devices and so much video, and that's just going to keep growing, so there's an imperative to adjust the architecture of networks today to provide services in a way that scales and that's very flexible for operators," said Bill Snow, vice president of engineering for ON.Lab.
In this pilot of an open-source, software-defined network operating system, ON.Lab is doing three things:
It is demonstrating that its software controllers work in a live deployment on a piece of the Internet2 network. It is learning what it can from this experience. It is seeing how its software-defined networking IP peering application works. Because software-defined networks operate differently than traditional IP networks, network operators need a way to bridge the two so they can communicate with each other and ease the inevitable transition to software-defined networks. That's the job of the peering application.
The peering application uses the Border Gateway Protocol (BGP) to exchange routing information with gateway hosts, which act like checkpoints on the Internet border between two networks. In this case, Internet2's software-defined network deploys the peering application to send traffic to the five university-run IP networks.
In this pilot of two technologies, Internet2 and the participating universities want to figure out the limitations of typical vendor approaches to software-defined networking and pressure them to make their technology more interoperable and programmable. They also want to find out what it will take to build a next-generation academic network.
Because higher education was a leader in advanced networking for so long, Boyd said, its engineers can play a powerful role in bringing back innovation to networking. And this pilot could kick-start the next wave of innovation.