多QoS约束的双目标最优的网格工作流调度研究  

Research on dual-objective optimal grid workflow scheduling with multiple QoS constraints

作　　者：李金忠[1] 夏洁武[1] 曾劲涛[1] 朱兵[1] 冷明[1] 

机构地区：[1]井冈山大学信息科学与传媒学院,江西吉安343009

出　　处：《计算机应用研究》2009年第9期3472-3474,共3页Application Research of Computers

基　　金：江西省教育厅科技项目(GJJ08417);吉安市科技局科研项目(科计字[2008]21号[1])

摘　　要：针对当前网格工作流调度算法中大多只考虑DAG结构的网格工作流,涉及QoS参数较少或将多QoS参数聚合成一个单目标函数进行优化调度,提出了一种多QoS约束的双目标最优的网格工作流调度算法。该算法是基于AGWL网格工作流模型和改进的MOPSO算法,其目标是在满足可靠性、可利用性和声誉这三维QoS参数约束下,同时最小化两个冲突目标,即响应时间和服务费用。通过与原MOPSO所设计的网格工作流调度算法比较,该算法能获得更优的优化解。Existing grid workflow scheduling algorithms commonly suffer by one or several of the following drawbacks ： most only considered grid workflow of DAG model, involved less QoS parameters or multidimensional QoS parameters would be aggre- gated into a single objective function for optimal scheduling. The paper presented an algorithm of dual-objective optimal grid workflow scheduling with multiple QoS constraints. The algorithm is based on AGWL grid workflow modeling and improved MOPSO algorithm, its goal is to simultaneously minimize two conflicting objectives- response time and service cost while mee- ting the three-dimensional QoS Constraints of reliability, availability and reputation. The proposed algorithm is compared with a grid workflow scheduling algorithm based on the original MOPSO algorithm, the experimental results show the better Pareto optincal solutions of algorithm.

关 键 词：服务质量 网格工作流 调度 多目标粒子群算法 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]