基于组件服务副本增删的云服务性能保障方法的研究

发布时间：2019-02-24 18:45

【摘要】：随着云计算的发展和服务需求的增多,出现了面向业务流程的云服务系统,即云环境下的基于组件服务的软件系统。系统中云服务以业务流形式提供服务,而业务流中的每个任务均被绑定在相应的云组件上,这些组件服务共同协作完成整个服务。每个组件服务可以有一个或多个副本并被部署在云环境资源池中。这里的云服务是一个服务业务流程,其具有组件服务并发请求变化,组件服务重要性差异,组件服务副本数目差异等特点,这些特点对云服务的性能产生一定影响并可能造成性能瓶颈、虚拟机利用率低、组件服务副本数目不足等问题。因此,为基于多副本的云服务制定有效的服务性能保障机制逐渐成为研究热点。在以往服务性能保障方法的基础上,本文借鉴服务组合中面向服务QoS的增量部署方法,分析了其不足之处,提出了一种基于组件服务副本增删的云服务性能保障方法。整个性能保障方法主要研究两个重点,即基于组件服务质量和重要性的云服务的瓶颈诊断方法和面向云服务性能保障的组件服务副本增删方法。对于瓶颈诊断,首先通过对影响云服务中每个组件服务的时延因素进行特征刻画与量化评估,得出每个组件服务质量,之后分析组件服务的调用关系和调用频率并根据PageRank算法计算出每个组件服务的重要性,最后综合组件服务质量和重要性得出云服务的瓶颈组件服务和最优组件服务；对于副本增删,首先判断需要副本增量部署还是副本删除,副本部署需要分析各组件的资源需求量和虚拟机SLA违背率,之后综合这两方面来确定适合部署副本的目标虚拟机,副本删除则需要分析部署了最优组件服务及副本的各虚拟机的资源使用情况。最后通过仿真实验验证了本文提出的基于组件服务质量和重要性的瓶颈诊断方法的准确性以及面向云服务性能保障的组件服务副本增删方法的有效性。此外,这种性能保障方法是以较少资源提供更优服务的初步研究,因此,本文研究的基于副本增删的云服务性能保障方法对后续的研究具有重要的参考意义。
[Abstract]:With the development of cloud computing and the increase of service demand, a business process-oriented cloud service system, that is, the component-based software system, has emerged in the cloud environment. Cloud services in the system provide services in the form of business flows, and each task in the business flow is bound to the corresponding cloud components, which work together to complete the whole service. Each component service can have one or more copies and be deployed in the cloud environment resource pool. The cloud service is a service business process, which has the characteristics of concurrent request change of component service, difference of importance of component service, number of replica of component service, etc. These characteristics have a certain impact on the performance of cloud services and may lead to performance bottlenecks, low utilization of virtual machines, insufficient number of replicas of component services, and so on. Therefore, the establishment of an effective service performance guarantee mechanism for cloud services based on multiple replicas has gradually become a research hotspot. On the basis of the previous service performance assurance methods, this paper uses the incremental deployment method of service-oriented QoS in service composition as reference, analyzes its shortcomings, and proposes a cloud service performance guarantee method based on component service replica addition and deletion. The whole performance assurance method mainly studies two key points, namely, the bottleneck diagnosis method of cloud service based on component quality of service and its importance, and the method of adding and deleting component service replica oriented to cloud service performance assurance. For bottleneck diagnosis, first of all, by characterizing and quantifying the delay factors that affect each component service in cloud service, the quality of service of each component is obtained. Then it analyzes the calling relation and frequency of component service and calculates the importance of each component service according to PageRank algorithm. Finally, the bottleneck component service and optimal component service of cloud service are obtained by synthesizing the quality and importance of component service. For replica additions and deletions, the first step is to determine whether the replica incremental deployment or the replica deletion is required. The replica deployment needs to analyze the resource requirements of each component and the virtual machine SLA violation rate, and then synthesizes these two aspects to determine the target virtual machine that is suitable for deploying the replica. Copy deletion requires an analysis of the resource usage of the virtual machines that deploy the optimal component service and replica. Finally, the accuracy of the proposed bottleneck diagnosis method based on the quality of service and importance of components is verified by simulation experiments, and the effectiveness of the component service replica adding and deleting method based on cloud service performance guarantee is also verified. In addition, this performance guarantee method is a preliminary study of providing better service with less resources. Therefore, the cloud service performance guarantee method based on replica addition and deletion is of great significance for future research.
【学位授予单位】：东北大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.09

【相似文献】