Percolation-based routing in the Internet

The uncontrollable growth of the Internet, breaking through meshing and multi-homing practices the existing topology-based prefix aggregation mechanisms, creates the necessity of revisiting some fundamental aspects in the inter-domain routing model due to severe scalability issues in routing table size. In this paper, we at first analyze the root causes of these problems and then exploit a promising solution based on on-demand routing and on a widely known uniform caching and searching algorithm. Such algorithm is based on bond percolation, a mathematical phase transition model well-suited for random walk searches in power law networks, automatically shielding nodes with limited connectivity from large traffic volumes and reducing the total traffic to scale sub-linearly with the network size. The proposed solution introduces limited modifications to the BGP protocol, ensuring backward compatibility and allowing gradual deployment throughout the Internet. It dramatically reduces the routing table size requirements in all the nodes participating to the search network while allowing reliable and efficient on-demand discovery of unknown routing information, as demonstrated through extensive simulation experiments.