SIGMETRICS '08 Proceedings of the 2008 ACM SIGMETRICS international conference on Measurement and modeling of computer systems
Proceedings of the 9th ACM SIGCOMM conference on Internet measurement conference
IEEE/ACM Transactions on Networking (TON)
Internet connectivity at the Autonomous System (AS)-level, defined in terms of pairwise business peering relationships, is constantly evolving. This evolution is largely a response to economic, political, and technological changes that impact the way ASs conduct their business. Existing research activities focusing on modeling AS-level Internet connectivity have been lacking in efforts to incorporate the very business-centric nature of ASs in establishing peering relationships and exchanging traffic with other ASs, and instead have been perusing generic graph-theoretic approaches. As a result, the fidelity of their findings and usefulness of their approaches from the networking perspective can sometimes be called into question. For one, we demonstrate that the widely used AS-level topology snapshots as exported from Border Gateway Protocol (BGP) routing data sets contain significantly incomplete connectivity information at the AS-level. This misleading picture of BGP-inferred AS-level Internet topology is due to routing policies of individual ASs, as a consequence of their business peering relationships. We also show that the generic graph growth model that takes the BGP-inferred AS graphs at face value fails to explain the evolution of the Internet connectivity at the AS-level. These findings suggest the need for an alternate approach that fully embraces the AS-specific semantics in explaining how the inter-AS connectivity is constructed and evolves over time. We present a new framework for modeling this AS-specific evolutionary process by identifying a set of criteria that ASs consider either in establishing a new peering relationship or in reassessing an existing relationship. The proposed framework is intended to capture key elements in the decision processes underlying these relationships. Taking into account the dual role of an AS as a customer and a peer, we propose a model that incorporates two distinct decision processes executed by an AS. We then populate the model with realistic inter-AS traffic demands derived from publicly available surrogate measurements. To reflect the enormous heterogeneity among customer or peer ASs, our models are also flexible enough to accommodate a wide range of AS-specific objectives. We demonstrate the potential of this new framework by considering different decision models in various realistic "what if" experiment scenarios. We implement these decision models to generate and study the evolution of the resulting AS graphs over time, and compare them against observed historical evolutionary features of the Internet at the AS level.