Poor visibility into the network hampers progress in a number of important research areas, from network troubleshooting to Internet topology and performance mapping. While the problem has been long recognized and has served as motivation for several efforts to build new or expand existing experimental platforms, capturing the edge of the Internet at sufficient scale remains an elusive goal. At its roots, the problem is one of incentives - in most of the commercial Internet the goals of those hosting the platform and the experimenters that use it are not the same. This effort explores a new model for Internet experimentation platforms that addresses the incentive problem by explicitly aligning the objectives of the experimenters with those of the users hosting the vantage points. The exploration is done in the context of Dasu, a system for Internet Service Provider characterization and Internet experimentation. The effort investigates (i) the challenges and opportunities offered by these types of platforms, (ii) how such platforms can support easy extensibility while remaining secure, and (iii) how to facilitate the design and deployment of experiments from the Internet edge while reducing the impact on both nodes? resources and the underlying network.
Broader Impact: The effort will offer the general Internet community with immediate benefits through a unique view of a user's ISP service levels. It will produce a set of algorithms, monitoring protocols, software systems, and tools that, by not requiring the deployment and maintenance of additional infrastructure, can immediately benefit their users, making a significant impact on society-at-large. The effort will provide the research community with access to an experimentation platform capable of capturing the diversity of the commercial Internet. It will make significant contributions toward a better understood, more transparent Internet and offer a much-needed testbed for evaluating the effectiveness of measurement techniques, tools and networked systems deployed at the edge of the Internet. Research findings and outcomes will be made available as publications, datasets and openly released software. The project will have a direct impact on education in communication, network and distributed systems.
