A load balance strategy for P2P networks

  • Authors:

  • Mitica Craus;Catalin Bulancea

  • Affiliations:

  • Department of Computer Science an Engineering, Technical University "Gh. Asachi", Iasi, Romania;Institute for Computer Science, Romanian Academy, Iasi, Romania

  • Venue:
  • ICCOMP'05 Proceedings of the 9th WSEAS International Conference on Computers
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

A new decentralized Load Balance (LB) paradigm for P2P Networks is presented and analyzed. Our strategy is based on a gradient map which is build by a pheromone-like technique inspired from the Ant Colony Optimization metaheuristic. The tasks are assumed to have a certain running time (T_run), which is considered to be a random value uniformly distributed in the interval (T_runmin, T_runmax) and a transfer time T_trans, which is also a random variable from the interval (T_transfmin, T_transfmax). New tasks can appear spontaneously. Each task is intending to wait a certain time in a workstation, time which cannot exceed a predefined value T_queue; the probability to request a transfer for another workstation is a random variable uniformly distributed in the interval (0,T_queue). T_queue parameter is depending of T_transf. When the new tasks do not appear in every processing node, LB is efficient.