Quantifying Internet End-to-End Route Similarity

Route similarity refers to the similarity of two routes between two nodes and an arbitrary third node. This intuitive concept plays an important role in distributed system deployment and path-edge inference. However, route similarity has not been quantitatively studied from an end-node perspective, and its properties are poorly understood. In this paper, we make an initial effort in quantifying route similarity by investigating a simple metric—RSIM. Using two large-scale traceroute data sets, we show that RSIM can be measured using only a small number of random traceroutes, and that it captures the similarity of both upstream and downstream routes. As a case study, we also describe how to use RSIM to infer shared path edges. We show that if a pair of end nodes have an RSIM value larger than 0.8, they have over 80% probability of sharing path edges.