Bloom Filter Based Inter-Domain Name Resolution: A Feasibility Study

The enormous size of the information space raises significant concerns regarding the scalability of name resolution in Information-Centric Networks, especially at a global scale. Recently, the use of Bloom filters has been proposed as a means to achieve a more compact representation of name resolution state. However, little attention has been paid to the expected performance or even the feasibility of such an approach. In this paper, we aim to fill this gap, presenting a feasibility study and performance analysis of Bloom filter based route-by-name inter-domain name resolution schemes. We propose a methodology for assessing the memory and processing resource requirements of the considered schemes and apply it on top of the inter-domain topology. Our investigation reveals that the skewed distribution of state across the inter-network results in a hard to balance trade-off between memory and processing resource requirements. We show that hardly any Bloom filter configuration i.e., size and bits-per-element, is able to reduce both types of resource requirements for all Autonomous Systems (ASes) in the network, while lowering resource requirements at one area of the inter-domain topology inflates resource requirements at another. Detailed simulations further show that the direct connection of multiple stub networks to tier-1 ASes, results in a dramatic increase of false positives, questioning the reliability of a BF-based inter-domain name resolution scheme.