论文总字数：26236字

摘 要

随着信息产业的飞速发展和互联网的普及，数据中心提供的信息服务平台将起着重要作用。快速发展的数据中心现在正面临着很多的问题, 其中基础设施资源的需求的快速上升，与数据中心单调高昂的资源提供方式产生了冲突,由此高成本、低效率、高耗能的传统数据中心成为眼下其最严峻的挑战之一。为达成资源的高效共享,用来应对大数据高速增长势头而产生的云计算,凭仗其更为灵活高效、低成本的运行方式成为新代数据中心重要的发展方向。而面向云环境的数据中心,目的在于大规模资源的整合。而有效的资源调度保障了新代数据中心部署大规模应用，它关系到数据中心的运营成本、整体性能以及可持续发展能力。

于是,本文着重研究面向云环境数据中心的高效资源调度。在此，我们将考虑多数据中心工作流调度问题，涉及了任务之间的分配顺序。我们将工作流中的任务分配给云（包括存储资源和计算资源），综合考虑任务分配给计算资源时的通信成本和计算成本，最小化工作流中任务完成所需的总的执行成本为优化目标。具体实现是在基于启发式的PSO工作流调度算法上，引入模拟退火机制改善粒子对应的解的质量，提高了处于或接近局部最优解的粒子在算法下一次迭代过程当中被选中的几率，从而可以有效地分配任务来节约成本。

关键词：云计算 工作流调度 粒子群优化算法 模拟退火

Experimental cloud computing platform and The reserch of Scheduling Algorithm for Discrete Optimization PSO

Abstract

With the rapid development of information industry and the wide popularization of the Internet data center as the foundation of information service platform has played an important role.The development of the data center is facing many problems now, which the demand of sharply rising infrastructure resources and data center the contradiction between single expensive way of resources, make the traditional data center of high cost, low efficiency, high energy consumption become one of the most serious challenges.In order to reach effective sharing of resources, the cloud computing which is purposed to deal with the high-speed growth of large data, with its more flexible ,efficacious and low-cost operation mode ，has become an important trend of the development of the next generation of data center.The data center which oriented the cloud, aimed at resource integration based on large-scale. Resource scheduling, as a important role of new generation of data center  in promoting the deployment of large scale, it is related to the data center operational costs, the overall performance and sustainable development capacity.

Therefore, this article focuses on efficient resource scheduling for the cloud data center. Here, we will consider more complicated workflow scheduling problem, which involves the order between tasks.Assign workflow tasks in the cloud (including computing and storage resources), and considering the resources allocated to the task of computational costs and communication costs between computing resources, in order to minimize the workflow task to complete the overall execution cost as the optimization goal.Here is based on the workflow scheduling heuristic PSO, on the basis of this,we add simulated annealing mechanism to improve the quality of the results of corresponding particles, increased at or close to the local optimal solution of the particle is selected in the process of the next iteration algorithm of probability, improve the efficiency of the PSO to assign task for cost savings

KEYWORDS: Cloud Computing；Workflow Scheduling； Particle Swarm Optimization Algorithm；Simulated Annealin

目 录

摘 要 I

Abstract II

第一章 绪论 5

1.1 研究背景和意义 5

1.2 研究现状 6

1.3 本论文研究内容 7

第二章 云计算的基础构架及核心技术 8

2.1云计算的概述 8

2.1.1云计算的简介 8

2.1.2 云计算基本原理 9

2.1.3 云计算基础构架 10

2.2 云模拟器介绍 12

2.3 云计算管理及任务调度的的分析 13

第三章 具有退火机制的PSO云计算工作流调度算法 14

3.1 工作流应用 14

3.2 模拟退火 14

3.2.1 模拟退火算法原理 14

3.2.2 模拟退火算法描述： 15

3.2.3 模拟退火算法的参数控制问题 16

3.3 粒子群优化算法 16

3.3.1 PSO算法的基本原理 16

3.3.2 PSO算法的性能与改进 17

3.4 工作流调度问题定义 18

3.5 改进PSO的算法 20

3.5.1 改进PSO算法的计算成本过程 20

3.5.2 改进PSO算法的自身计算过程 20

第四章 云计算环境中工作流调度算法的嵌入与实验 22

4.1 开发环境 22

4.2 DataCenter控制器中改进PSO算法的嵌入 22

4.3 模拟实验配置 23

4.3.1 基于改进PSO算法的任务调度 23

4.3.2 程序设计及体现 26

4.3.3 实验结果分析 27

4.3.4 变动计算资源成本 27

4.3.5 改进PSO算法的收敛 28

第五章 总结与展望 29

5.1 全文总结 29

5.2下阶段研究方向展望 29

结束语 31

参考文献 32

致 谢 35

第一章 绪论

1.1 研究背景和意义

请支付后下载全文，论文总字数：26236字