基于模拟退火遗传算法的云资源调度方法-华东理工大学学报.PDF

DOI ：10.14135/ki.1006 -3080.20180416001 基于模拟退火遗传算法的云资源调度方法 1,2 1 1 高天阳 , 虞慧群 , 范贵生 （1.华东理工大学计算机科学与工程系, 上海 200237 ；2.上海市计算机软件评测重点实验室, 上海 201112） 摘要: 代理云是为用户从不同的云提供商中发现和挑选合适的云计算服务。随着应用系统的规模和复 杂度的增加，如何选择最优的云服务，并在多个云服务提供商分散部署应用，有效地缓解供应商锁定问题 成为代理云所面临的难题。本文提出了一种基于模拟退火遗传算法的云资源调度方法，主要解决在代理云 系统上搜索满足应用QoS 需求资源的问题。实验结果表明，本文算法相比传统遗传算法具有较快的收敛速 度，在不影响解精度的前提下，提高了算法效率。 关键词:代理云；遗传算法；模拟退火算法；服务质量 中图分类号:TP393 文献标志:A Simulated Annealing Genetic Algorithm Based Approach to Cloud Resource Scheduling 1,2 1 1 GAO Tian-y ang , YU Hui-qun , FAN Gui-sheng (1.Department of Computer Science and Engineering, East China University of Science and Technology, Shanghai 200237, China; 2.Shanghai Key Laboratory of Computer Software Evaluating and Testing, Shanghai 201112, China) Abstract: Cloud brokers aim to help consumers to discover and select suitable cloud computing ser- vices from a variety of different cloud providers. With the increase in the scale and complexity of application systems, how to select the optimal cloud services and deploy them in multiple cloud provider s to effectively mitigate the problem of vendor lock-in becomes a challenge for cloud brokers . Genetic algorithm and simu- lated annealing algorithm are two possible candidates for the optimization purpose. Genetic algorithm has a strong global search ability, but the algorithm is easy to fall into the local optimal solution. Simulated an- nealing algorithm has strong local search ability and c