Comparing the performance and efficiency of two popular DHTs in interpersonal communication

Distributed Hash Tables (DHTs) have been deployed in a wide variety of peer-to-peer (P2P) networks. While the traditional use of P2P technology is focused around content sharing, the current trend is in applying the P2P technology for new application areas, such as interpersonal communications. Despite active research and standardization efforts, such as the development of Peer-to-Peer Session Initiation Protocol (P2PSIP), there has been little effort to evaluate the suitability of different DHTs for interpersonal communications. As there are significant differences between the requirements of various application types, the task of selecting an appropriate DHT algorithm for a particular application type is not trivial. Therefore an evaluation is needed. In this paper, we compare the performance and efficiency of two popular DHT algorithms, Kademlia and Chord, in P2PSIP environment. We are particularly interested in their messaging overhead and routing performance in different sizes of overlay networks. The results reveal that in identical application scenarios, Kademlia outperforms Chord by shorter lookup path lengths, whereas Chord scales better to larger networks in the terms of messaging cost.