On multilevel voting

Summary In a distributed system mutual exclusion is often used to maintain consistency when restricted operations are performed. Mechanisms guaranteeing mutual exclusions should be both resilient and efficient. Resiliency implies high resource availability in the face of failures, while efficiency implies low overhead incurred by performing restricted operations. In this paper, we propose and study a general paradigm, called multilevel voting, which provides a general framework to assist in the design of resilient and efficient mutual exclusion mechanisms. The proposed method uses multiple level quorum consensus. Unlike another method based on the use of multiple quorum consensus, the proposed model only contains one type of integrity constraints. This has the benefit of being conceptually simple and easy to reason about. The strong resemblance with the traditional quorum consensus makes it easy for the proposed paradigm to embed any technique based on traditional quorum schemes. We show that the proposed model represents the exact class of coteries. This means that not only does it have all the power of coteries, but also all schemes under the model are correct. Thus, should the need arise, we can interchange two schemes freely without using any extra mechanisms to ensure correctness. We study a number of issues that have impact on performance such as the degree of a multilevel scheme and the order of a coterie. We explain how the model can be extended also to model the schemes for the synchronization of read and write of replicated data. We provide algorithms for the design of multilevel schemes in the context of mutual exclusion and that of read and write of replicated data.J. Tang received the M.S. degree in computer science from the University of Iowa in 1983 and the Ph.D degree in computer science from the Pennsylvania State University in 1988. Since 1988, he has been an assistant professor with the Department of Computer Science, Memorial University of Newfoundland, St. John's, Newfoundland, Canada. His current research interests include distributed computing, fault tolerance in distributed systems, database modeling and multidatabase systems.This work was supported in part by the Natural Sciences and Engineering Research Council of Canada, individual operating grant OGP0041916