基于服务粒度空间的多租户服务组合方法研究

发布时间：2018-05-18 20:47

本文选题：云计算 + 多租户　；参考：《山东大学》2014年硕士论文

【摘要】：近年来,云计算得到了学术界和工业界的广泛关注,作为一种新颖的软件提供方式,软件即服务(SaaS)的模式正变得日渐流行。随着用户个性化多样化需求越来越多,面向多租户的网络应用是未来应用交付的主要形式。为了适应复杂应用开发,企业中的应用组件也向着模块化、细粒度的方向发展。从而使得新的应用构建可以通过组件的按需组合来实现。因此,面向多租户的服务组合是企业灵活构建应用的关键。然而,随着云计算的发展,租户的个性化和多样化需求日益强烈,传统的服务组合方法已不再适用,使得多租户服务组合面临诸多挑战。主要表现为：由于用户需求的多样化,有的需求很抽象有的很具体,而呈现出多粒度性；SaaS模式下多租户的特点使得几乎每个租户对应用的需求都不一样,定制的需求便得日趋强烈；多租户的定制开发会带来其他服务与现有服务之间的关联关系；由于业务需求的变化,服务总是频繁地从一个版本演化到另一个版本而呈现出的频繁的演化。为了解决上述挑战,本文提出了一个服务粒度空间模型,为多租户服务组合提供了语义基础。该服务粒度空间支持层次化,继承,演化的关联关系以及版本化的特点,有效应对云计算环境下多租户服务组合的各种挑战。一方面,该服务粒度空间能够将海量无序的服务组织成层次化有序的结构,另一方面,利用该服务粒度空间能够容易高效地开发多租户SaaS应用。另外,本文将所提出的方法用到基于SaaS的社保服务系统的开发,实现了一个多租户服务组合原型系统。本文的主要工作和贡献包括： 1)通过研究多租户服务组合中存在的多粒度性,广泛的定制,由定制引发的关联以及频繁的演化特点,提出了一个支持层次化、继承、演化的关联关系以及版本化的服务粒度空间。该服务粒度空间不仅可以为服务组合提供语义基础,而且可以有效应对云计算环境下构建SaaS应用时的多租户服务组合的挑战。 2)通过研究服务聚簇以及关联挖掘技术,提出了构建服务粒度空间的方法并给出了实现个性化和演化的策略。该方法将海量无序的服务组织成有序的结构,能够快速构建多租户应用。 3)通过研究所提出的服务粒度空间特有的层次化特点,提出了基于服务粒度空间的多租户服务组合方案生成算法,在满足租户QoS需求的前提下保证得到多粒度的服务组合方案,实现租户的多样化需求。 4)由于服务空间规模巨大,在对服务组合方案中每个抽象服务选择绑定具体服务时耗时过长,严重影响了多租户服务组合的性能,为此提出了基于skyline计算的两层服务空间裁剪的算法,使得服务组合性能大大提高。本文基于SaaS的社保应用构建进行了简单的原型系统的设计和实现,对租户的个性化定制操作进行介绍；对提出的服务粒度空间的构建,服务空间裁剪策略以及基于服务粒度空间的多租户服务组合方法进行了验证。
[Abstract]:In recent years, cloud computing has attracted wide attention from academic and industrial circles. As a new way of providing software, the model of software and service (SaaS) is becoming increasingly popular. With the increasing demand for personalized diversification of users, multi tenant oriented network applications are the main form of failure to apply delivery. In order to adapt to complex applications, In the development, the application components in the enterprise also develop in the direction of modularization and fine-grained. Thus, the new application construction can be realized through the combination of components. Therefore, the key to the flexible construction of the application is the multi tenant oriented service composition. However, with the development of cloud computing, the individual and diversified demands of the tenants are increasingly strong. The traditional service composition method is no longer applicable, making the multi tenant service combination face many challenges. The main performance is: the diversification of the user needs, some of the requirements very abstract and very specific, and presents a multi granularity; the characteristics of multi tenants in the SaaS model make it different from each tenant to the application. The demand is increasingly strong; the custom development of multi tenants will bring about the relationship between other services and existing services, and the frequent evolution of services from one version to another because of the change in business requirements.
In order to solve these challenges, this paper presents a service granularity spatial model, which provides a semantic basis for multi tenant service composition. The service granularity space supports hierarchical, inheriting, evolutionary Association and versioning characteristics. There are various challenges to the multi tenant service portfolio in the cloud computing environment. On the one hand, the service granularity is granularity. Space can organize massive disordered services into hierarchical and orderly structure, on the other hand, the use of the service granularity space can easily and efficiently develop multi tenant SaaS applications. In addition, the proposed method is used in the development of social security service system based on SaaS, and a multi tenant service combination prototype system has been realized.
The main work and contributions of this article include:
1) by studying the multi granularity, extensive customization, the association and the frequent evolution characteristics of the multi tenant service composition, a hierarchical, inherit, evolutionary correlation and versioning service granularity space is proposed. The service granularity space can not only provide semantic basis for service composition, but also the service granularity space. It has the effect of challenging the multi tenant service composition of SaaS applications in cloud computing environment.
2) through the study of service clustering and association mining technology, the method of building the service granularity space is proposed and the strategy of implementing personalization and evolution is given. This method will organize a large number of disordered services into an orderly structure, and can quickly construct multi tenant applications.
3) through the study of the special hierarchical characteristics of service granularity space proposed in this study, a multi tenant service combination scheme generation algorithm based on service granularity space is proposed, which guarantees the multi granularity of service portfolio to meet the tenant's QoS requirements, and realizes the diversity requirement of the tenants.
4) because of the large scale of service space, it takes too long time for each abstract service to bind specific services in the service composition scheme, which seriously affects the performance of the multi tenant service composition. Therefore, a two layer service space clipping algorithm based on Skyline computing is proposed, which greatly improves the performance of the service combination.
In this paper, the design and implementation of a simple prototype system based on the social security application of SaaS is designed and implemented, and the personalized custom operation of the tenant is introduced. The construction of service granularity space, the service space clipping strategy and the multi tenant service grouping method based on service granularity space are verified.
【学位授予单位】：山东大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.09

【参考文献】