一种多属性网格任务调度算法及仿真研究

为使网格任务调度时能更多地考虑任务和资源之间的各种属性.通过对常用静态调度算法的分析,吸收了Min-min和Max-min等算法的思想,将影响网格任务调度的诸多属性归纳为两类因素.提出了一种针对多属性任务的调度算法MASA,经过截断处理、归一化、加权计算等方法得出任务与资源之间的匹配矩阵,指导任务调度.仿真实验测试结果表明,在相同任务和资源环境下,通过改变不同属性的加权系数能得到所期望的调度结果,使具有高加权系数属性的任务在调度时更具优势.此算法具有灵活性,属性可增可减,能根据具体情况进行配置,以满足具体应用需求.

A grid task scheduling algorithm based on multi-attribute and its simulation

In order to consider multi-attribute between grid tasks and computing resources, through analysis of the commonly used static scheduling algorithm, this paper absorbed the idea of the Min-min and Max-rain algorithm. The properties are grouped into two categories of factors affect grid task scheduling. After truncation processing, normalized, a weighted method it calculates the matching matrix and guides task scheduling. The simulation test results show that change the properties of the weighting factor can be the results of scheduling at environment of the same tasks and resources. Properties of task with a high weighting factor has an advantage. This algorithm has the flexibility, property can increase and decrease, can be configured to meet specific application requirements under the circumstances. Key words