成本时间限制下的网格分类调度算法研究  

Grid Classified Optimization Scheduling Algorithm under Limitation of Cost and Time

作　　者：朱春玲[1] 唐小勇[2] 李肯立[2] 

机构地区：[1]华中科技大学计算机学院,湖北武汉430074 [2]湖南大学计算机与通信学院,湖南长沙410082

出　　处：《计算机应用研究》2007年第1期31-33,共3页Application Research of Computers

基　　金：国家自然科学基金资助项目(60273075);教育部重点资助项目(05128)

摘　　要：在网格环境中,由于资源广域分布、异构、动态且有多个管理域,调度一组具有多QoS需求如成本、时间的独立任务是一个非常重要的问题。针对网格任务的成本和执行时间要求,提出了一种基于网格经济模型,根据实际执行成本和预算成本进行分类的网格分类优化调度算法。模拟实际网格任务调度实验表明,该算法能很好地满足网格环境中不同用户的需求。The problem of scheduling a set of independent tasks （ a meta-task） with multiple QoS needs, such as time and cost, is a very important problem in grid environment because the resources are geographic distributed, heterogeneous, dynamic and owned by various organize domain. According to the cost requirement and the demand of executing time of grid task, a classified optimization-scheduling algorithm is introduced. This algorithm can well trade-off the actual executing cost and its budget in grid economy model. Simulated experiments in virtual grid environment show that it can satisfy the demand of different grid users.

关 键 词：网格经济模型 调度 优化 MAKESPAN 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]