"I Can't Get No Satisfaction": Helping Autonomous Systems Identify Their Unsatisfied Interdomain Interests

Given the distributed and business-driven nature of the Internet, economic interests of Autonomous Systems (ASes) may be incompatible. Previous works studied specific effects of incompatible interests, especially BGP policy conflicts leading to routing and forwarding anomalies. In this paper, we rather focus on the effects of incompatible interests that do not trigger such anomalies. We take the perspective of a single AS: We show that incompatible interests can have a tangible impact on its business, and provide a classification of its unsatisfied interests. Since incompatible interests cannot be solved automatically, our effort is directed to support network managers in their business decisions. Hence, we describe algorithms to identify and assess their impact, as well as a prototype of a warning system aimed at signaling the most relevant unsatisfied interests. We evaluate our prototype on real data from two operational networks. In addition to illustrate the potential of our system, our evaluation shows that unsatisfied interest are relatively frequent and likely affect a significant amount of traffic in practice.