网格环境下的高效动态任务调度算法

High-Efficiency Dynamic Task-Scheduling Algorithm for Grids

网格系统具有异构性、动态性和分布性,这使得网格中的任务调度变得十分复杂,要求调度算法具有动态性和自适应性.文中将群体智能技术引入网格的任务调度中,针对一组相互独立的任务调度问题,提出了一种新的动态任务调度算法.该算法利用蜂群与环境的交互模型来实现网格中动态的任务分配;同时根据蜂群的自组织社会层次的概念,解决了算法中出现的竞争问题.实验结果表明,相比于现有的方法,该算法对于网格中的动态环境具有更好的适应性,且调度性能更优.

The effective task scheduling in a grid environment is very difficult due to the dynamic requirements, various loads and heterogeneous distributed resources, etc. of the system. So it is necessary to research a dynamic and adaptive algorithm of task scheduling. In this paper, the swarm intelligence is introduced into the task scheduling in a grid environment, and a novel dynamic task-scheduling algorithm for a group of independent tasks is proposed, where the model describing the interaction between the wasp colony and the environment is used to implement the dynamic task scheduling in grids, and the contest problem is solved according to the self-organized dominance hierarchy of a wasp colony. Experimental results show that, compared with the existing methods, the proposed algorithm is more adaptive to the dynamic grid environment, and possesses better scheduling performance.