ABSTRACT:

Internet is facing many challenges that cannot be solved easily through ad-hoc patches. To address these challenges, many research programs and projects have been initiated and many solutions are being proposed. However, before we have a new architecture that can motivate the Internet Service Providers (ISPs) to deploy and evolve, we need to address two issues: (1) know the current status better by appropriately evaluating the existing Internet, and (2) find how various incentives and strategies will affect the deployment of the new architecture. For (1), we define a series of quantitative metrics that can potentially unify results from several measurement projects using different approaches, and can be an intrinsic part of the future Internet architecture for monitoring and evaluation. Using these metrics, we systematically evaluate the current inter-domain routing system and reveal many "Autonomous System (AS)-level" observations and key lessons for new Internet architectures. Particularly, the evaluation results reveal the imbalance underlying the inter-domain routing system and how the deployment of future Internet architectures can benefit from these findings. With these findings, for (2), appropriate deployment strategies of the future architecture changes can be formed with balanced incentives for both customers and ISPs. The results can be used to shape the short-term and long-term goals for new architectures that are simple evolutions of the current Internet (so called dirty slate architectures) and to some extent to clean-slate architectures.

Complete paper in Adobe Acrobat format.