Internet topology on AS-level: model, generation methods and tool

Joanna Tomasik (1) , Marc-Antoine Weisser (1)

1 E3S - Supélec Sciences des Systèmes

Numerous studies based on complex measurement platforms have been carried out for over ten years now in order to discover the Internet topology on domain level. It turns out that this topology exhibits certain invariant properties such as a distribution of node degree. This distribution follows a power law. Moreover, the revealed topology is hierarchical. The hierarchy is caused by commercial contracts signed between domain operators. The routing realized by Border Gateway Protocol (BGP) is strongly influenced by these commercial relationships because operators do not want to make public the routes they know, as announcing certain routes would deprive them of a possible financial benefit. Consequently, routes available in BGP tables are of particular shape (valley-free). This fact has an impact on the performance of protocols which are proposed notably to assure the Quality of Service (QoS). In order to evaluate the performance of new protocols in the inter-domain context their designers have to have at their disposal a random topology generator which is able to furnish a random graph whose nodes' degree follows a power law typical for the Internet, and to impose the commercial hierarchy on it. Our aSHIIP (autonomous Supélec Hierarchy Inter-domain Program) does both: its synthetic topologies are realistic and the hierarchy, which it introduces, corresponds to the one of the Internet. After explaining the reasons for our study, we present the methods which we propose to use for the flat Internet topology generation which satisfies the realism of the Internet. Next, we explain our algorithm used to induct the commercial hierarchy. The algorithm is heuristic because, as we prove in this paper, the underlying problem is NP-complete. We then evaluate the topologies generated with aSHIIP. The result is a reliable flat Internet-like topology generator which also allows the modeler to introduce the realistic hierarchy. We are convinced we can recommend it to modelers dealing with performance evaluation of protocols for the Internet on domain level.

Mots clés: random topology generation; hierarchy on AS-level

Domaines: Modélisation et simulation; Réseaux et télécommunications [cs.NI]

Joanna Tomasik, Marc-Antoine Weisser. Internet topology on AS-level: model, generation methods and tool. The 29th IEEE International Performance Computing and Communications Conference (IPCCC 2010), Dec 2010, Albuquerque, New Mexico, United States. 8 p. ⟨hal-00546721⟩