ON CONSISTENCY OF DATA IN STRUCTURED OVERLAY NETWORKS ∗ Tallat M. Shafaat † , Monika Moser ‡ , Ali Ghodsi ∤ , Thorsten Sch¨ utt ‡ , Seif Haridi † , Alexander Reinefeld ‡ Abstract Data consistency can be violated in Distributed Hash Tables (DHTs) due to in- consistent lookups. In this paper, we identify the events leading to inconsistent lookups and inconsistent responsibilities for a key. We find the inaccuracy of failure detectors as the main reason for inconsistencies. By simulations with inaccurate failure detectors, we study the probability of reaching a system con- figuration which may lead to inconsistent data. We analyze majority-based algo- rithms for operations on replicated data. To ensure that concurrent operations do not violate consistency, they have to use non-disjoint sets of replicas. We analyt- ically derive the probability of concurrent operations including disjoint replica sets. By combining the simulation and analytical results, we show that the prob- ability for a violation of data consistency is negligibly low for majority-based algorithms in DHTs. 1. Introduction Peer-to-peer systems have gained tremendous popularity in recent years due to characteristics of scalability, fault-tolerance and self-management. Struc- tured Overlay Networks (SONs) are a major class of these peer-to-peer system, examples of SONs include Chord [3], Chord# [6], Pastry [10] and DKS [7]. SONs provide lookup services for Internet-scale applications. Distributed Hash Tables (DHTs) use a SON’s lookup service to provide a put/get interface for distributed systems with eventual consistency guarantees [11]. In contrast, many distributed systems require stronger consistency guarantees, relying on ∗ This research work is carried out under the SELFMAN project funded by the European Commission and the Network of Excellence CoreGRID funded by the European Commission. †Royal Institute of Technology (KTH), {tallat , haridi}(at )kth.se ‡Zuse Institute Berlin (ZIB), {moser , schuett , ar}(at )zib.de ∤ Swedish Institute of Computer Science (SICS), ali(at )sics.se