On the Marginal Utility of Deploying Measurement Infrastructure

An emerging strategy to gain insight into the conditions and con gurations in the core of the Internet is the use of end-to-end measurements from a set of distributed measurement points. As the number of measurement points increases, so does the number of paths through the network that can be studied, which is an argument for a \more is better" approach to wide-area measurement. However, basic questions about the utility of increasing the number of measurements and measurement sites have not yet been addressed. In this paper, we present a more re ned and quanti able understanding of the marginal utility of performing wide-area measurements. We focus on problems in Internet topology discovery, namely, discovering the set of nodes and links which comprise the Internet backbone, discovering the degree distribution of these nodes, and classifying nodes according to their role. We provide a model for how the discovery process scales as the number of measurements and measurement sites increase and validate this model against a set of traceroutes run across the Internet from 17 measurement sites run to 60,000 destinations. We characterize the topology in terms of nodes, links, node degree distribution, and distribution of end-to-end ows using statistical and information-theoretic techniques. We show that the utility of adding sources beyond the second source goes down signi cantly from the perspective of interface, node, link and node degree discovery. We show that the utility of adding destinations is constant for interfaces, nodes, links and node degree indicating that it is more important to add destinations than sources.