Adaptive Control of Stable Load Balancing Algorithm for Parallel Cluster Computing
Qingyang Meng, Jianzhong Qiao, Jun Liu, Sukuan Lin · 2009
In this paper, the linear and nonlinear models are put forward first to describe load balancing system for parallel cluster computing. Through analyzing the stability of linear model, this paper finds the asymptotic stable condition of load balancing system with some assumptions, and the simulation results are given to compare the results in theory later. Based on the analyzing and simulation results, an adaptive load balancing algorithm(ALBA) which considers the delay, execution time and scalability of system is presented to enhance the performance of parallel computing. At last, the simulation for ALBA is given to prove the practicability of the load balancing algorithm.