云环境下基于蚁群算法的资源调度策略研究

发布时间：2018-04-17 17:09

本文选题：云计算 + 资源调度　；参考：《广东工业大学》2014年硕士论文

【摘要】：云计算是一种由分布式计算、网格计算以及并行计算演变而来的新型计算模式。其主要运用虚拟化技术,将云端数据中心的各种资源虚拟化成为资源池进行管理以及对外提供服务,并且形成对用户的“按需分配、按量计费”的商业模式。这些资源对用户而言是透明的,用户只需要知道云端数据中心所提供的服务并选择自己需要的服务,不需要知道任务的具体执行过程与具体执行位置,云端把最终结果返回给用户。云计算商业潜力巨大,对未来IT运营模式影响深远,如今已成为国内外企业以及研究机构研究的热点。由于云计算环境的异构性、自治性、动态性以及云环境中使用了虚拟化技术,因此云计算环境中的资源分配方式跟以往的分布式计算、并行计算以及网格计算大不相同。为了适应云端数据中心规模的扩大和用户以及任务数量的不断增加,云环境下资源调度的目的在于提出一种优化的资源调度策略,使得数据中心中的虚拟机资源能够满足用户提出的Qos要求的同时又能实现资源的合理高效利用。在简要分析了云计算以及云环境下资源调度的研究现状,总结了现有的资源调度策略的优缺点以及改进方向,并介绍云计算与云环境下资源调度策略的相关概念以及技术体系的基础上,本文主要做了以下三个方面的工作：第一,分析、研究传统的蚁群算法在云计算环境下的资源调度上存在的问题,包括时间跨度大、负载均衡度较低以及优化目标单一等：第二,在详细介绍蚁群算法、模拟退火算法的基本思想、特点以及设计要素的基础上,针对云计算编程最常用的Map/Reduce的框架,设计出一种新算法蚁群模拟退火算法(ACOSA),该算法融合蚁群算法以及模拟退火算法,以最小化调度时间为主要目标,引入了任务与资源的匹配因子和负载均衡度,利用蚁群算法得到一组任务到资源的优化解,然后通过模拟退火算法,对解进行路径的优化和信息素的更新,最后得到全局最优解；第三：在论文的最后,详细地介绍了云计算仿真平台CloudSim,对其进行重新编译,实现了提出的本论文提出的ACOSA算法,通过与基于原始蚁群算法的云环境资源调度策略相比较,验证了本文提出的调度策略在时间跨度以及负载均衡方面有良好的表现。
[Abstract]:Cloud computing is a new computing model evolved from distributed computing, grid computing and parallel computing.It mainly uses the virtualization technology to virtualize all kinds of resources of the cloud data center into the resource pool to manage and provide the service to the outside, and form the business mode of "according to demand, according to the quantity charge" to the user.These resources are transparent to users, who only need to know the services provided by the cloud data center and choose the services they need, without knowing the specific execution process and location of the task.The cloud returns the final result to the user.Cloud computing has great commercial potential and has a profound impact on the future IT business model. Now cloud computing has become a hot research topic for enterprises and research institutions at home and abroad.Because of the heterogeneity, autonomy, dynamics and virtualization technology used in cloud computing environment, resource allocation in cloud computing environment is very different from distributed computing, parallel computing and grid computing.In order to adapt to the expansion of the scale of cloud data center and the increasing number of users and tasks, the purpose of resource scheduling in cloud environment is to propose an optimized resource scheduling strategy.The virtual machine resources in the data center can meet the Qos requirements of users and realize the rational and efficient utilization of the resources at the same time.In this paper, the current situation of research on cloud computing and resource scheduling in cloud environment is briefly analyzed, and the advantages and disadvantages of existing resource scheduling strategies are summarized as well as the direction of improvement.Based on the introduction of cloud computing and resource scheduling strategy in cloud environment, this paper mainly does the following three aspects of work: first, analysis,This paper studies the problems of traditional ant colony algorithm in resource scheduling in cloud computing environment, including long time span, low load balance and single optimization goal. Secondly, the ant colony algorithm is introduced in detail.Based on the basic idea, characteristics and design elements of simulated annealing algorithm, a new ant colony simulated annealing algorithm (ACOSA) is designed for the framework of Map/Reduce, which is the most commonly used framework of cloud computing programming. The algorithm combines ant colony algorithm and simulated annealing algorithm.In order to minimize the scheduling time, the matching factor and load balancing degree of task and resource are introduced, and a set of optimal solution from task to resource is obtained by using ant colony algorithm, and then simulated annealing algorithm is used to solve the problem.Finally, the global optimal solution is obtained by optimizing the path of the solution and updating the pheromone. Thirdly, at the end of the paper, the cloud computing simulation platform CloudSims is introduced in detail, which is recompiled, and the ACOSA algorithm proposed in this paper is realized.Compared with the original ant colony algorithm based resource scheduling strategy in cloud environment, the proposed scheduling strategy has good performance in time span and load balancing.
【学位授予单位】：广东工业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP18;TP393.01

【参考文献】