企业私有云计算平台资源分配研究与设计

发布时间：2018-05-16 11:01

本文选题：云计算 + 私有云　；参考：《湖南大学》2012年硕士论文

【摘要】：随着企业IT系统的日益庞大和复杂化,服务器硬件资源的浪费越来越严重。实践表明,企业的应用系统对硬件资源的使用在时间上是不均等的,而传统的方案通常都必须按照应用系统峰值性能要求来采购硬件资源,否则在高负载时,用户体验将会受到严重影响。但是在空闲时段,它们所使用的服务器资源几乎下降到10%左右。企业私有云可以作为对企业IT应用系统硬件需求的快速响应来应对这一问题。私有云平台通过虚拟机的方式提供资源,在高峰时段可以动态的添加资源而在低负载时回收资源。本文通过对大量文献和实际企业环境的研究,讨论了标准企业私有云的架构和建立的方法,分析了企业私有云计算平台实现的重点和难点。研究表明,为了使企业私有云计算平台能稳定可靠的运行,且能提供足够的负载能力和可扩展性,需要建立有效的方法进行性能需求的计算。基于这一需求和特点,本文对动态资源分配机制、性能指标选择和基线确认等关键问题进行了分析,给出了选择有效性能指标并测量准确基线和通过虚拟机池及时响应资源需求变化的方法,通过设计和建立实验模型证明了资源动态分配的可行性。本文通过一系列的实验和测试数据以及对国内外流行的企业私有云建设方案的对比研究,深入论证了建立一套符合企业自身实际情况的私有云所需要准备的必要条件以及建设方式。通过设计一套实验模型阐明和验证了动态资源分配的实现方式和效果。通过加入和企业应用相关的性能指标,完善错误检测和边界检查机制等方法对这套模型进行了扩展,使其具有更多动态资源分配的高级功能。
[Abstract]:With the increasing complexity of enterprise IT system, the waste of server hardware resources is becoming more and more serious. The practice shows that the use of hardware resources in enterprise application systems is not equal in time, but the traditional schemes usually have to purchase hardware resources according to the peak performance requirements of the application system, otherwise, when the load is high, The user experience will be severely affected. But in idle time, the server resources they use almost drop to about 10%. The enterprise private cloud can respond to this problem as a quick response to the hardware requirements of enterprise IT applications. Private cloud platform provides resources through virtual machines and can dynamically add resources during peak hours and recycle resources at low load. This paper discusses the architecture and method of establishing standard enterprise private cloud, and analyzes the key points and difficulties in the implementation of enterprise private cloud computing platform through the research of a large number of documents and actual enterprise environment. The research shows that in order to make the enterprise private cloud computing platform stable and reliable, and to provide sufficient load capacity and scalability, it is necessary to establish an effective method to calculate the performance requirements. Based on this requirement and characteristics, this paper analyzes the key issues, such as dynamic resource allocation mechanism, performance index selection and baseline validation, etc. The methods of selecting effective performance indexes and measuring accurate baselines and responding to resource demand changes in time through virtual machine pool are given. The feasibility of dynamic resource allocation is proved by designing and establishing experimental model. Through a series of experimental and test data and a comparative study of the popular enterprise private cloud construction schemes at home and abroad, The necessary conditions and construction methods for the establishment of a private cloud in accordance with the actual situation of the enterprise are discussed in depth. An experimental model is designed to illustrate and verify the realization and effect of dynamic resource allocation. By adding the performance index related to enterprise application and perfecting the mechanism of error detection and boundary check, the model is extended to have more advanced functions of dynamic resource allocation.
【学位授予单位】：湖南大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP302

【参考文献】