多数据中心负载均衡调度的研究
发布时间：2018-10-29 12:47
【摘要】：云计算为数据中心带来了机遇，也带来了挑战。云数据中心的资源需要得到合理的调度才能更高效地向用户提供服务。物理服务器为了保持较高的性能，必须将CPU、内存等的利用率保持在某个水平以下，因而需要均衡物理机之间的负载，尽量减少某些机器过载而另外一些机器空转的情况。多数据中心环境下，数据中心要得到整体上较高的服务水平，也要尽量保持较低的资源利用率，因而在资源的数量一定时，均衡各数据中心之间的负载，有助于提高服务水平。 本文提出了多数据中心环境下云资源共享模型，并遵照该模型实现了云资源共享系统，详细介绍了系统的设计和实现，包括系统主模块设计、各模块的通信机制、系统信息的保存方法等。之后，本文根据该系统所调度任务的考虑生命周期和资源共享的特点，提出了应用于该模型的多数据中心负载均衡机制，包括离线和在线两种方式的负载均衡算法。 最后，本文使用实验模拟的方式将所提出的算法同轮询、随机、最小负载优先、最长处理时间优先等经典负载均衡算法进行对比，验证所提出算法的性能。算法对比的指标包括用户的虚拟机任务请求拒绝个数、多个数据中心的整体不均衡度、最大的数据中心综合利用率、最小的数据中心利用率与最大的数据中心综合利用率之比等。实验数据显示，本文所提出的算法与其它算法相比，在上述指标方面有明显的优势。
[Abstract]:Cloud computing brings both opportunities and challenges to data centers. The resources of cloud data center need to be scheduled reasonably to provide service to users more efficiently. In order to maintain high performance, the physical server must keep the utilization of CPU, memory below a certain level, so it is necessary to balance the load between the physical machines and minimize the overload of some machines and the idling of others. In the multi-data center environment, the data center should get a higher service level as a whole, but also maintain a low resource utilization rate. Therefore, when the number of resources is fixed, balancing the load between the data centers will help to improve the service level. In this paper, a cloud resource sharing model in multi-data center environment is proposed, and the cloud resource sharing system is implemented according to the model. The design and implementation of the system are introduced in detail, including the design of the main module of the system, the communication mechanism of each module. Methods of saving system information. Then, according to the characteristics of life cycle and resource sharing, this paper proposes a multi-data center load balancing mechanism for this model, including offline and online load balancing algorithms. Finally, the proposed algorithm is compared with the classical load balancing algorithms, such as polling, random, minimum load first and maximum processing time priority, in order to verify the performance of the proposed algorithm. The indexes of algorithm comparison include the number of requests rejected by the user's virtual machine task, the overall imbalance of multiple data centers, and the maximum comprehensive utilization ratio of data centers. The ratio of minimum data center utilization to maximum data center comprehensive utilization. Experimental data show that the proposed algorithm has obvious advantages over other algorithms.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP308
本文编号：2297735
[Abstract]:Cloud computing brings both opportunities and challenges to data centers. The resources of cloud data center need to be scheduled reasonably to provide service to users more efficiently. In order to maintain high performance, the physical server must keep the utilization of CPU, memory below a certain level, so it is necessary to balance the load between the physical machines and minimize the overload of some machines and the idling of others. In the multi-data center environment, the data center should get a higher service level as a whole, but also maintain a low resource utilization rate. Therefore, when the number of resources is fixed, balancing the load between the data centers will help to improve the service level. In this paper, a cloud resource sharing model in multi-data center environment is proposed, and the cloud resource sharing system is implemented according to the model. The design and implementation of the system are introduced in detail, including the design of the main module of the system, the communication mechanism of each module. Methods of saving system information. Then, according to the characteristics of life cycle and resource sharing, this paper proposes a multi-data center load balancing mechanism for this model, including offline and online load balancing algorithms. Finally, the proposed algorithm is compared with the classical load balancing algorithms, such as polling, random, minimum load first and maximum processing time priority, in order to verify the performance of the proposed algorithm. The indexes of algorithm comparison include the number of requests rejected by the user's virtual machine task, the overall imbalance of multiple data centers, and the maximum comprehensive utilization ratio of data centers. The ratio of minimum data center utilization to maximum data center comprehensive utilization. Experimental data show that the proposed algorithm has obvious advantages over other algorithms.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP308
【参考文献】
中国期刊全文数据库 前2条
1 赵雷霆;;运营商级云计算数据中心发展研究[J];信息安全与技术;2011年08期
2 周集良,彭小宁,王正华;基于集群的负载平衡调度算法研究与实现[J];计算机工程;2005年12期
中国博士学位论文全文数据库 前4条
1 刘晓茜;云计算数据中心结构及其调度机制研究[D];中国科学技术大学;2011年
2 章文嵩;可伸缩网络服务的研究与实现[D];中国人民解放军国防科学技术大学;2000年
3 王智明;云数据中心资源调度机制研究[D];北京邮电大学;2012年
4 孙鑫;面向云环境数据中心的高效资源调度机制研究[D];北京邮电大学;2012年
，本文编号：2297735
本文链接：https://www.wllwen.com/kejilunwen/jisuanjikexuelunwen/2297735.html
