基于个性化需求的金融云平台宿主机选择算法的设计与实现

发布时间：2018-07-10 07:34

本文选题：云计算 + 宿主机选择　；参考：《吉林大学》2016年硕士论文

【摘要】：近几年,云计算概念越来越热,云,是一种全新的服务模式,云平台通过网络和虚拟化技术对基础硬件进行整合,使用户脱离了对IT设施及平台的管理,将更多的精力放到业务逻辑上,同时还提高了计算资源的使用率,一改以往传统IT的服务方式,让IT服务从使用到维护均跨越了时空的限制。IT领域的每一次变革势必会波及到金融领域,云计算给金融领域带来了新的发展契机和新的需求。各家金融机构纷纷利用云计算这个平台对大数据中心进行升级改造,以拓展新的业务领域,优化管理工作。这就对云计算提出了更高的要求,不仅要求保证金融业务和数据的安全性,同时要求技术力量具有高的可靠性。一个好的云架构应该为金融业务提供充分的可靠性和安全性。云计算服务是建立在部署于宿主机机群上的虚拟资源上的。传统的虚拟机部署工作通常单一地依据宿主机当前CPU状况来作为目标宿主,宿主机其他的资源状况未作为部署参数考虑进来,同时也忽略了虚拟机所要消耗的资源特点。不同用户完成的业务目标不同,因此提出的资源请求也具有不同特点。单纯地依靠宿主机CPU这一指标作为部署依据,可能导致宿主机不能满足用户的IT资源请求,随着云平台的用户增加,进而可能影响整个云平台的负载均衡,导致IT资源得不到有效合理的利用。因此,宿主机的选择、虚拟资源的部署对于云计算提供优质的服务有至关重要的作用。本文提出了基于个性化需求的宿主机选择模型,该模型将用户的资源请求量化,再根据云中宿主机的性能情况与量化结果进行比对,选择最适合用户的宿主机予以部署虚拟资源。同时还对该模型中涉及的负载预测机制、负载均衡机制均进行了深入的研究,并提出了适合该模型的负载预测机制和负载均衡机制。通过实验验证:与传统的宿主机选择方法相比,该模型在充分满足用户个性化需求的前提下,在降低虚拟机迁移频率、均衡云平台的负载,保证系统稳定性方面有明显提高。
[Abstract]:In recent years, cloud computing concept is becoming more and more hot, cloud is a new service model, cloud platform through network and virtualization technology to integrate the basic hardware, so that users out of the management of IT facilities and platforms, By focusing more on business logic and increasing the utilization rate of computing resources, we have changed the traditional IT service style. Every revolution in IT field will inevitably spread to the financial field. Cloud computing has brought new development opportunities and new needs to the financial field. Financial institutions use cloud computing as a platform to upgrade and upgrade the big data Center to expand new areas of business and optimize management. This puts forward higher requirements for cloud computing, not only to ensure the security of financial services and data, but also to require high reliability of technical forces. A good cloud architecture should provide adequate reliability and security for financial operations. Cloud computing services are built on virtual resources deployed on a host cluster. The traditional virtual machine deployment is usually based solely on the current CPU status of the host as the target host. The other resource conditions of the host are not considered as deployment parameters and the characteristics of the resources consumed by the virtual machine are ignored. Different users accomplish different business objectives, so the resource request has different characteristics. Relying solely on the host CPU as the basis for deployment may result in the host being unable to meet the user's IT resource requests. As the number of users on the cloud platform increases, it may affect the load balance of the entire cloud platform. As a result, IT resources can not be used effectively and reasonably. Therefore, the choice of host and the deployment of virtual resources are very important for cloud computing to provide high quality services. In this paper, a host selection model based on personalized requirements is proposed. The model quantifies users' resource requests, and then compares them with the quantization results according to the performance of the host in the cloud. Select the most suitable host for the user to deploy the virtual resource. At the same time, the load forecasting mechanism and load balancing mechanism involved in the model are deeply studied, and the load forecasting mechanism and load balancing mechanism suitable for the model are proposed. The experimental results show that compared with the traditional host selection method, the model can reduce the migration frequency of virtual machine, balance the load of cloud platform, and ensure the stability of the system.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP393.09

【参考文献】