Nets-NBD: Intelligent routing control systems: panacea or imminent disaster?

Georgia Tech Research Corporation, Atlanta GA

Investigators

Abstract

Recently, both the networking industry and Internet research communities have shown significant interest in ``Intelligent Routing Control'' (IRC) systems. There are two types of IRC systems: routing overlays and intelligent multihoming. First, IRC systems perform load-sensitive routing, allowing dynamic path changes based on the measured performance of candidate paths. Second, with IRC systems it becomes possible to route different classes of traffic to the same destination through different paths. Third, IRC systems give edge networks the flexibility to choose their path to common destinations. This project investigates IRC systems from a critical perspective, examining both their potential to improve network routing and application performance, but also the risk that such systems may introduce in the Internet in the form of instability and conflicting routing incentives. The project's main research tasks are to: investigate both routing overlays and intelligent multihoming and compare the pros and cons of the two architectures; identify the key issues in the design and operation of IRC systems; design algorithms and mechanisms for the stable and efficient operation of IRC Systems; evaluate the previous results through realistic simulations and wide-scale experiments over Planetlab and other testbeds. The final outcome of this project will be a coherent set of designs and algorithms for IRC systems, considering both performance and stability, and focusing not only on the incentives of individual players but also on the global utility of the Internet as a whole. The project can have a significant impact on the design and deployment of IRC systems. The project's results will be disseminated through research publications, and through direct communication with manufacturers of IRC systems and network operators.