融合容错需求和资源约束的云容错服务适配方法研究

发布时间：2018-03-22 07:44

本文选题：云计算　切入点：容错即服务　出处：《内蒙古大学》2017年硕士论文　论文类型：学位论文

【摘要】：云计算以其虚拟化、按需服务等特点吸引了越来越多的应用系统迁移到云上。但是,云平台底层基础设施的高度复杂性使得云数据中心会经受大量的故障,并降低云应用系统的可靠性。因此,容错对云应用系统变得十分重要。如果由云基础架构服务提供商为云应用系统提供容错服务,不同云应用系统的容错需求将不能被充分顾及;相反,如果由云应用系统自行实现容错机制,云基础架构底层物理资源将很难得到充分利用。因此,容错即服务模式应运而生,即由第三方容错服务提供商以容错即服务的方式为云应用系统提供容错服务。这不仅能够顾全不同云应用系统的容错需求,还能充分利用底层的云资源。在目前的容错即服务研究中,很多研究都假设用户在定义容错需求时指定了容错方法,但这会存在如下不足:首先,容错解决的主要问题是提高应用的可靠性,而可靠性是用户最关心的问题。让用户指定容错方法会使用户的容错需求无法直接反映用户关心的可靠性。其次,如果用户指定容错方法,那么满足用户容错需求所需的资源将是固定的。云容错服务供应商现有资源的固定性以及满足用户容错需求所需资源的固定性,使供应商只能为固定数量的用户提供容错服务,这会使云资源得不到充分利用,进而对供应商的收益产生影响。因此,本文针对用户容错需求定义无法直接反映容错可靠性以及云容错服务供应商资源得不到充分利用等问题,提出了一种融合容错需求和资源约束的云容错服务适配方法。对用户的容错需求进行详细的定义,并从云容错服务供应商的角度,分别在其资源是否足够以及是否考虑容错方法间的切换开销的场景下,研究容错服务适配方法并使用最优化理论生成该适配方法下的容错服务。本文的工作主要包括以下几个部分:给出用户容错需求的合理描述方式;计算使用不同的容错方法满足用户可靠性需求时的资源需求、成本开销和响应时间,并对容错方法的容错效果进行分析;计算容错方法间的切换开销;提出不同场景下的容错服务适配方法并生成相应的容错服务。实验结果表明,本文提出的容错服务适配方法生成的容错服务,不仅能够满足用户的容错需求,还能使云容错服务供应商通过充分利用资源获得最大收益。
[Abstract]:Cloud computing has attracted more and more applications to migrate to the cloud because of its virtualization and on-demand services. However, the high complexity of the underlying infrastructure of the cloud platform makes cloud data centers subject to a large number of failures. And reduce the reliability of cloud applications. Therefore, fault tolerance becomes very important for cloud applications. If cloud infrastructure service providers provide fault-tolerant services for cloud applications, The fault-tolerant requirements of different cloud applications will not be fully taken into account; on the contrary, if cloud applications implement fault tolerance mechanisms themselves, it will be difficult to make full use of the underlying physical resources of cloud infrastructure. That is, the third party fault-tolerant service providers provide fault-tolerant services to cloud applications in a fault-tolerant manner, which not only can cater for the fault-tolerant requirements of different cloud applications, In the current research on fault-tolerant as services, many studies assume that users specify fault-tolerant methods when defining fault-tolerant requirements, but this will have the following shortcomings: first, The main problem to be solved by fault tolerance is to improve the reliability of the application, and reliability is the most important issue for the user. To let the user specify the fault-tolerant method will not directly reflect the reliability of the user's concern. If the user specifies a fault-tolerant method, the resources required to meet the user's fault-tolerant requirements will be fixed. A vendor can only provide fault-tolerant services to a fixed number of users, which causes cloud resources to be underutilized, which in turn has an impact on the vendor's revenue. In this paper, the definition of user fault tolerance requirements can not directly reflect the fault tolerance reliability and cloud fault-tolerant service provider resources are not fully utilized and so on. A cloud fault-tolerant service adaptation method combining fault-tolerant requirements and resource constraints is proposed. The user's fault-tolerant requirements are defined in detail, and from the point of view of cloud fault-tolerant service providers, In scenarios where its resources are adequate and whether the switching overhead between fault-tolerant methods is considered, The fault-tolerant service adaptation method is studied and the fault-tolerant service under the adaptive method is generated by the optimization theory. The main work of this paper includes the following parts: the reasonable description method of the user's fault-tolerant requirement is given; The resource requirements, cost overhead and response time of different fault-tolerant methods are calculated, and the fault-tolerant effect of the fault-tolerant methods is analyzed, and the switching overhead between the fault-tolerant methods is calculated. The fault tolerant service adaptation method in different scenarios is proposed and the corresponding fault-tolerant service is generated. Experimental results show that the fault-tolerant service generated by the proposed fault-tolerant service adaptation method can not only meet the fault-tolerant needs of users. It also enables cloud fault-tolerant service providers to maximize benefits by making full use of resources.
【学位授予单位】：内蒙古大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP302.8

【参考文献】