The cost of concurrent, low-contention Read&Modify&Write

Costas Busch, Marios Mavronicolas, Paul Spirakis

Research output: Contribution to journalConference article

4 Scopus citations

Abstract

The possibility or impossibility and the corresponding costs of devising concurrent, low-contention implementations of atomic Read&Modify&Write (or RMW) operations in a distributed system were addressed. A natural class of monotone RMW operations associated with monotone groups was introduced. A certain class of algebraic groups was also presented. A Monotone Linearizability Lemma was proved and employed as a chief combinatorial instrument. It establishes inherent ordering constraints of linearizability for a certain class of executions of any distributed system implementing a monotone RMW operation. A lower bound on latency for switching networks that implement monotone groups was proved by using Monotone Linearizability Lemma.

Original languageEnglish (US)
Pages (from-to)373-400
Number of pages28
JournalTheoretical Computer Science
Volume333
Issue number3
DOIs
StatePublished - Mar 3 2005
Externally publishedYes
EventStructural Information and Communication Complexity - Umea, Sweden
Duration: Jun 18 2003Jun 20 2003

Keywords

  • Distributed computing
  • Linearizability
  • Lower bounds
  • Monotone Linearizability Lemma
  • Switching networks
  • Synchronization

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint Dive into the research topics of 'The cost of concurrent, low-contention Read&Modify&Write'. Together they form a unique fingerprint.

  • Cite this