Applied Metarouting Project

The bedrock service of the Internet is connectivity --- allowing the exchange of data between hosts located in diverse, autonomously administered networks. Internet connectivity is implemented using dynamic routing protocolsthat exchange reachability information between tens of thousands of network routers. Current routing protocols are few in number and are not well suited for many networks. As a result, the existing protocols are pressed into service in highly complex and contorted ways, thus limiting the flexibility of network design, increasing the cost of network operations, and lowering the level of network robustness.We refer to this as the routing protocol ossification problem. Griffin and Sobrinho recently proposed metarouting as a potential solution to this problem. Metarouting is based on the design of a metalanguage for thespecification of routing protocols. Implementing such a metalanguage on a router would enable network operators to define, implement, and deploy new protocols that are tailored to the needs of their networks. It would also allow network researchers to experiment in a rich space of new routing protocols. One innovation of the metarouting paper was to show how a metalanguage can be defined so that protocol specifications could be automatically validated to guarantee both protocol convergence and loop-free forwarding. Lack of correctness in the current interdomain routing protocol (BGP) results in routing anomalies that are very difficult to debug. There is a large gap between the theoretical sketch of metarouting and usable, efficient implementations. This Case for Support is aimed at closing that gap. We seek funding for the design, development, and validation of the first prototype implementation of metarouting. This is a challenging goal that will require solutions to many difficult problems, including the design of a user-oriented metalanguage, compilation of metalanguage specifications to efficientimplementations, and extracting generic algorithms from current routing protocol implementations. Our implementation will build on the existing code base of XORP (www.xorp.org), a suite of open source routing protocols developed in a project led by Dr Mark Handley of UCL.