A preliminary study on ancestral voting algorithm for availability improvement of mutual exclusion in partitioned distributed systems

  • Authors:
  • Faraneh Zarafshan;Abbas Karimi;S. A. R. Al-Haddad;M. Iqbal Saripan;Shamala Subramaniam

  • Affiliations:
  • Department of Computer and Communication Systems Engineering, Faculty of Engineering, UPM, Malayasia and Department of Computer Engineering, Faculty of Engineering, IAU-Arak Branch, Iran;Department of Computer Engineering, Faculty of Engineering, IAU-Arak Branch, Iran;Department of Computer and Communication Systems Engineering, Faculty of Engineering, UPM, Malayasia;Department of Computer and Communication Systems Engineering, Faculty of Engineering, UPM, Malayasia;Department of Communication Technology & Networks, Faculty of Computer Science and IT, UPM, Malayasia

  • Venue:
  • ICCC'11 Proceedings of the 2011 international conference on Computers and computing
  • Year:
  • 2011

Quantified Score

Hi-index 0.00

Visualization

Abstract

Mutual exclusion as one of the main problems associated with distributed systems requires that at any time t, only one site can access the common copies of file f. As a result of node or link failures, distributed system may be partitioned into two or more groups of nodes which cannot communicate each other anymore; meanwhile they must be able to update their copies of file f consistently, in order to keep the system up. Voting algorithms are the main solutions presented to maintain mutual exclusion in case of distributed system partitioning. So far several techniques have been presented to increase the availability of mutual exclusion in partitioned distributed systems. In this paper a new dynamic voting approach called as ancestral voting is presented to increase the availability of distributed system. Our analyses show that this algorithm can maintain mutual exclusion and increase the availability of distributed system between 2.86% and 59.83% in comparison with previous algorithms. Furthermore, it is very simple and does not need many calculations.