基于动力学演化算法的云任务与虚拟机分配策略研究  

Cloud Task and Virtual Machine Allocation Strategy Research Based on Dynamics Evolutionary Algorithm

作　　者：徐星[1] 魏波[2,3] 应伟勤 李元香[2,3] 

机构地区：[1]景德镇陶瓷学院信息工程学院,景德镇333000 [2]武汉大学软件工程国家重点实验室 [3]计算机学院,武汉430072 [4]华南理工大学软件学院,广州510006

出　　处：《科学技术与工程》2013年第1期85-89,共5页Science Technology and Engineering

基　　金：国家科技支撑计划(2012BAH25F02);国家自然科学基金(61070009);国家青年科学基金(61202313、122);江西省自然(青年)科学基金项目(20122BAB211036、20121BAB201044);江西省教育厅青年基金项目(GJJ12514);景德镇市科技局项目(2011-1-47);广东省自然科学基金(S2011040002472);高等学校博士学科点专项科研基金(20110172120035);中央高校基本科研业务费专项资金(2011ZM0107)资助

摘　　要：建立了云任务和虚拟机分配的数学模型,并将其转换成一个组合优化问题,利用改进的演化算法进行求解。算法中设计了两种编码方案,并根据这两种编码表示设计了杂交和变异算子,并引入了动力学选择机制让种群中的个体都有机会参与演化。采用了两组实验数据进行测试,并和随机分配算法、顺序分配算法以及贪心分配算法进行比较。实验结果表明该演化算法获得的结果最优。The mathematical model of the cloud tasks and the virtual machines allocation strategy were estab- lished and it was converted into a combination optimization problem. An improved evolutionary algorithm was ap- plied to solve this problem. The two encoding schemes were designed, and then the mutation operator and the crossover operator were also designed in accordance with the encoding schemes. The dynamics selection mechanism was introduced into the algorithm and it allowed all individuals of the population having the opportunity to partici- pate in the evolution. Two groups of experimental data were used to the performance of the evolutionary algorithm, and the evolutionary algorithm was compared with the random allocation algorithm, the sequential allocation algo- rithm and the greedy allocation algorithm. The experimental results show that the evolutionary algorithm obtains the best results.

关 键 词：演化算法 云计算 分配策略 动力学 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]