EDonkey & eMule's Kad: Measurements & attacks

Abstract

This article reports on the results of our measurement study of the Kad network. Although several fully decentralized peer-to-peer systems have been proposed in the literature, most existing systems still employ a centralized architecture. The Kad network is a notable exception. Since the demise of the Overnet network, the Kad network has become the most popular peer-to-peer system based on a distributed hash table. It is likely that its user base will continue to grow in numbers over the next few years due to the system's scalability and reliability. The contribution of the article is twofold. First, we compare the two networks accessed by eMule: the centralized paradigmof the eDonkey network and the structured, distributed approach pursued by the Kad network. We re-engineer the eDonkey server software and integrate two modified servers into the eDonkey network in order to monitor traffic. Additionally, we implement a Kad client exploiting a design weakness to spy on the traffic at arbitrary locations in the ID space. The collected data provides insights into the spacial and temporal distributions of the peers' activity. Moreover, it allows us to study the searched content. The article also discusses problems related to the collection of such data sets and investigates techniques to verify the representativeness of the measured data. Second, this article shows that today's Kad network can be attacked in several ways. Our simple attacks could be used either to hamper the correct functioning of the network itself, to censor content, or to harm other entities in the Internet not participating in the Kad network, such as ordinary web servers. While there are heuristics to improve the robustness of Kad, we believe that the attacks cannot be thwarted easily in a fully decentralized peer-to-peer system, i.e., without some kind of a centralized certification and verification authority. This result may be relevant in the context of the current debate on the design of a clean-slate network architecture for the Internet which is based on concepts known from the peer-to-peer paradigm.