Distributed Computing

Attiya, Hagit; Welch, Jennifer L.

doi:10.1002/0471478210

Cited by 608 publications

(319 citation statements)

References 0 publications

Supporting

Mentioning

314

Contrasting

Unclassified

Order By: Relevance

“…The algorithm implementing Alpha (r, v) using a shared memory is described in Figure 2 (it is close to algorithms that can be found in [10,11]). A simple examination of its code shows that it is a wait-free algorithm: if p i does not crash while executing Alpha (r, v) it will terminate (at line 3, 8 or 9) whatever the behavior of the other processes.…”

Section: The Algorithmmentioning

confidence: 97%

The Alpha of Indulgent Consensus

Guerraoui¹,

Raynal²

2006

The Computer Journal

View full text Add to dashboard Cite

This paper presents a simple framework unifying a family of consensus algorithms that can tolerate process crash failures and asynchronous periods of the network, also called indulgent consensus algorithms. Key to the framework is a new abstraction we introduce here, called Alpha, and which precisely captures consensus safety. Implementations of Alpha in shared memory, storage area network, message passing and active disk systems are presented, leading to directly derived consensus algorithms suited to these communication media. The paper also considers the case where the number of processes is unknown and can be arbitrarily large.

show abstract

Section: The Algorithmmentioning

confidence: 97%

The Alpha of Indulgent Consensus

Guerraoui¹,

Raynal²

2006

The Computer Journal

View full text Add to dashboard Cite

show abstract

“…This yields substantial improvements in the best known running times of algorithms for information spreading, in particular, on graphs that have small conductance but large weak conductance, from polynomial to polylogarithmic number of rounds. 2 Since the best known running times of algorithms for full information spreading inversely depend on the conductance, which may be small due to communication bottlenecks, a natural direction towards speeding up information spreading is to identify such bottlenecks and choose these links with higher probability, compared to other neighboring links. However, detecting bottlenecks does not seem easy.…”

Section: φC(g)mentioning

confidence: 99%

Fast Information Spreading in Graphs with Large Weak Conductance

Censor-Hillel¹,

Shachnai²

2012

SIAM J. Comput.

View full text Add to dashboard Cite

Gathering data from nodes in a network is at the heart of many distributed applications, most notably, while performing a global task. We consider information spreading among n nodes of a network, where each node v has a message m(v) which must be received by all other nodes. The time required for information spreading has been previously upper-bounded with an inverse relationship to the conductance of the underlying communication graph. This implies high running times for graphs with small conductance.The main contribution of this paper is an information spreading algorithm which overcomes communication bottlenecks and thus achieves fast information spreading for a wide class of graphs, despite their small conductance. As a key tool in our study we use the recently defined concept of weak conductance, a generalization of classic graph conductance which measures how well-connected the components of a graph are. Our hybrid algorithm, which alternates between random and deterministic communication phases, exploits the connectivity within components by first applying partial information spreading, after which messages are sent across bottlenecks, thus spreading further throughout the network. This yields substantial improvements over the best known running times of algorithms for information spreading on any graph that has a large weak conductance, from polynomial to polylogarithmic number of rounds.We demonstrate the power of fast information spreading in accomplishing global tasks on the leader election problem, which lies at the core of distributed computing. Our results yield an algorithm for leader election that has a scalable running time on graphs with large weak conductance, improving significantly upon previous results.

show abstract

“…Consensus is a classic problem in distributed computing, mostly studied assuming that processes communicate by shared variables or through message passing networks [4,21]. Most of the literature on consensus concerns the presence of processor faults, that can be either crash or Byzantine, starting from the seminal paper [22]; see the recent book [23] for a comprehensive survey.…”

Section: Related Workmentioning

confidence: 99%

Communication Complexity of Consensus in Anonymous Message Passing Systems

Fusco

2011

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. We consider the message complexity of achieving consensus in synchronous anonymous message passing systems. Unlabeled processors (nodes) communicate through links of a network. In each round every processor can exchange messages with all neighbors and the duration of each transmission is one round. An adversary wakes up some subset of processors at possibly different times and assigns them arbitrary numerical input values. All other processors are dormant and do not have input values. Any message wakes up a dormant processor. The goal of consensus is to wake up all processors and have them agree on one of the input values. We seek deterministic consensus algorithms using as few messages as possible. As opposed to most of the literature on consensus, the difficulty of our scenario are not faults (we assume that the network is fault-free) but the arbitrary network topology combined with the anonymity of nodes. For unknown n-node networks we show a consensus algorithm using O(n 2 ) messages; this complexity is optimal for this class. We show that if the network is known, then the complexity of consensus decreases significantly. Our main contribution is an algorithm that uses O(n 3/2 log 2 n) messages on any n-node network and we show that some networks require Ω(n log n) messages to achieve consensus. We also observe that availability of distinct labels of nodes helps to improve complexity of consensus for known networks but has no effect for the class of unknown networks. Indeed, even with labeled nodes, Ω(n 2 ) messages are sometimes necessary if the network is unknown but for known labeled networks consensus can be always achieved with O(n) messages.

show abstract

Distributed Computing

Cited by 608 publications

References 0 publications

The Alpha of Indulgent Consensus

The Alpha of Indulgent Consensus

Fast Information Spreading in Graphs with Large Weak Conductance

Communication Complexity of Consensus in Anonymous Message Passing Systems

Contact Info

Product

Resources

About