Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network

  Advanced Search  

Small Byzantine Quorum Systems JeanPhilippe Martin, Lorenzo Alvisi, Michael Dahlin

Summary: Small Byzantine Quorum Systems
Jean­Philippe Martin, Lorenzo Alvisi, Michael Dahlin
University of Texas at Austin ­ Dept. of Computer Science
Email: fjpmartin, lorenzo, dahling@cs.utexas.edu
In this paper we present two protocols for asynchronous Byzantine Quorum Systems (BQS) built on top of reliable
channels---one for self­verifying data and the other for any data. Our protocols tolerate f Byzantine failures with f
fewer servers than existing solutions by eliminating nonessential work in the write protocol and by using read and
write quorums of different sizes. In practice, however, engineering asynchronous reliable channels is difficult in many
environments. To address this concern, we modify the original asynchronous BQS protocol of Malkhi and Reiter to
work on unreliable channels and discuss how our two new asynchronous protocols can be used to derive an efficient
protocol for synchronous Byzantine systems.
1. Introduction
Quorum systems are valuable tools for implementing highly available distributed shared memory. The
principle behind their use is that if a shared variable is stored at a set of servers, then read and write operations
need only be performed at some set of servers (a quorum). The intersection property of quorums ensures that
each read has access to the most recently written value of the variable. Any practical use of quorum systems
must account for the possibility that some of the servers may be faulty; hence, quorum systems must enforce
the intersection property even in the presence of failures. Mahlki and Reiter introduce quorum systems,
called masking quorum systems, that guarantee data availability in the presence of arbitrary (Byzantine)


Source: Alvisi, Lorenzo - Department of Computer Sciences, University of Texas at Austin


Collections: Computer Technologies and Information Sciences