AMHERST, Mass. – The National Science Foundation announced recently that computer scientist Arun Venkataramani and computer engineer Lixin Gao at the University of Massachusetts Amherst have received a three-year, $1.2 million grant to develop and test a method to improve interdomain routing, “the top-level protocol that holds the entire Internet together,” they explain. New routing strategies should improve fundamental robustness, security and manageability and “benefit anyone who relies on the Internet today,” the researchers add.
Right now, Venkataramani explains, an interdomain or between-networks protocol called Border Gateway Protocol (BGP) is the protocol that all service providers use to route Internet traffic. He adds, “It’s the language spoken by all 60,000+ different networks active on the Internet today and there is no choice, it’s the only one.”
But BGP is prone to a number of fundamental problems such as “intermittent routing unavailability, security problems, management headaches and most importantly, difficulty evolving to meet these challenges,” Gao and Venkataramani point out.
This means that users can encounter long pauses during searches and websites can be unreachable. Also with BGP, network managers have great difficulty finding out where the problem originates. Gao says, “This problem of long delays is not the only one, but it’s a major one, and it can cost a company, losing it customers. This happens all the time because today’s interdomain routing is also fragile. It’s vulnerable and events can bring down huge portions of the entire Internet. It happens regularly. One reason is there is no single entity that has a big picture of what is going on, no manager,” she adds.
To address the situation, the researchers and their graduate students Xiaozhe Shao and Shahrooz Pouryousef, whose research has laid groundwork for the new approach, propose using a logically centralized interdomain routing architecture they have dubbed CIRCA. It will shift management of Internet traffic to the cloud, Venkataramani explains.
He adds that while communication by servers on the ground might take hundreds of milliseconds, in the cloud the same operation may take only one millisecond from one machine to another. “It’s orders of magnitude faster, and in the cloud we can easily afford more bandwidth resources, too. The photons have less distance to travel in the cloud than on the ground. All these factors make outsourcing the decision-making to the cloud more advantageous.” The researchers say this new approach of enabling interdomain routing as a service is “a radically different approach compared to today’s practice.”
Gao says their project will also take advantage of a very popular and widely embraced approach of software-defined networking (SDN) and adapt it from use in single networks to include the entire Internet.
“Inspired by SDN, we will deploy it through the whole Internet while respecting the confidentiality of the proprietary information of each network’s routing policies. There are security and privacy issues in doing this,” she adds. The researchers expect their cloud-based system will be more secure than the Internet is today because “you can react faster to events like an attack or a power failure,” Gao notes.
Shifting interdomain traffic control to the cloud to avoid routers on the ground and “heavy duty switching,” Gao says, plus SDN principles and policy routing, all contribute to the promise of better networking in a logically centralized system, the researchers say.
Venkataramani adds, “As an analogy, it’s a little bit like using live turn-by-turn navigation to tell you where to go as opposed to you reading a map or asking around for directions. One feels efficient and easier than the other. It’s a similar idea in interdomain routing, a server in the cloud is giving real-time feedback about how to proceed to packets on the Internet. The traffic still has to go through the routers on the ground, but the decision about how they will get to their respective destinations is made at a higher level in the cloud much more efficiently.”
The scientists plan to develop prototypes as proof of concept that their approach will work when implemented in real situations, they explain. Venkataramani says, “In addition to theory, we want to make sure the system converges faster and that we get the improvements we expect. We won’t change the way people do things overnight, but if it works well, we hope people will start to incrementally deploy this. The whole thing is collaborative.” Gao adds, “We expect only some might adopt it right away, but they’ll see the benefit and others may follow through.”