Analysis of Link Lifetimes and Neighbor Selection in Switching DHTs

Abstract

Abstract-Several models of user churn, resilience, and link lifetime have recently appeared in the literature [13], [14], [36], [37]; however, these results do not directly apply to classical Distributed Hash Tables (DHTs) in which neighbor replacement occurs not only when current users die, but also when new users arrive into the system, and where replacement choices are often restricted to the successor of the failed zone in the DHT space. To understand neighbor churn in such networks, which we call switching DHTs, this paper proposes a simple, yet accurate, model for capturing link dynamics in structured P2P systems and obtains the distribution of link lifetimes for fairly generic DHTs. Similar to [9], our results show that deterministic networks (e.g., Chord [30], CAN [25]) unfortunately do not extract much benefit from heavy-tailed user lifetimes since link durations are dominated by small remaining lifetimes of newly arriving users that replace the more reliable existing neighbors. We also examine link lifetimes in randomized DHTs equipped with multiple choices for each link and show that selecting the best neighbor in these scenarios is rather complicated as it depends on the desired load-balancing, link resilience, and overhead. We offer insight into the various selection algorithms, their performance, and possibilities for improvement.

Unstructured P2P link lifetimes redux

“…This indicates that there exists an inherent tradeoff between in and out-edge resilience. As V becomes stochastically larger, W gets stochastically smaller and eventually converges in distribution to L. While a somewhat similar result was observed in DHTs [6], [24], the reasons for these phenomena are completely different as we discuss below.…”

“…Therefore, p(x) reduces message overhead and increases resilience of out-links at the expense of lowering resilience of in-links. In the worst case, F W may "deteriorate" down to F L , which is reminiscent of the situation occurring in DHTs [24], where it happens due to the arrival of new users who take over the zones of existing neighbors.…”

“…Many P2P properties (e.g., message overhead, resilience to disconnection, and ability to reach other peers with queries) depend on the behavior of node degree, which is determined solely by the lifetime of edges in the graph. Despite the sizeable volume of analytical work on P2P networks (e.g., [4], [9], [10], [11], [12], [13], [15], [20], [23], [25], [24], [26]), accurate characterization of link lifetime has been elusive.…”

On the effect of peer online times on the lookup service of Chord and Kademlia P2P systems

“…In addition to the aforementioned works, there exist some papers [21] [22] where the effects of long-lived peers are analyzed.…”

“…Although there are several performance evaluations in the literature [15]- [20], up to the authors' best knowledge, none of these works analyzes the effects of long-lived peers on both DHTs. However, there exist some works that address issues related to long-lived peers [21], [22].…”