摘要
提出了一种新的网格任务调度模式,针对网格计算资源有组织、松耦合、自治等特性,建立基于多层次虚拟组织形式的计算资源模型;根据网格环境中应用任务粗粒度、特定资源依赖等特点,建立了网格任务的描述模型;提出并实现了相应的子任务生成算法、任务初始调度算法及自动调整算法。设计实现了能够支持仿真及实际网格计算环境可扩展网格任务调度器,通过理论分析和仿真实验对算法的正确性、效果和效率进行了评价。
This paper introduced a novel task scheduling pattern for grid computing. The grid resource model was based on multi-layers of virtual organization, reflecting that the autonomic grid computing resources were organized but loose coupled. The description model of grid task was based on the main feathers of most grid applications such as rough granularity and dependended on specific resources. Based on the models, designed algorithms for subtask generation algorithm, task initial scheduling and auto-adjust. According to the task scheduling pattern, designed and implemented an extensible scheduler supporting both simulation and actual grid computing environment. The evaluation on validity, effect and performance of the algorithm was from the results theoretical analysis and simulation.
出处
《计算机应用研究》
CSCD
北大核心
2008年第5期1500-1503,1510,共5页
Application Research of Computers
基金
国家自然科学基金重大研究计划资助项目(90412007)
关键词
网格计算
任务调度
任务调度模式
算法
调度器
仿真
grid computing
task scheduling
task scheduling pattern
algorithm
scheduler
simulation
