【作者】 江丽；

【导师】 李培峰； 朱巧明；

【作者基本信息】 苏州大学 ， 计算机应用技术， 2013， 硕士

【摘要】 随着云计算商业模式的飞速发展，越来越多的软件企业进入SaaS（Software as aService，软件即服务）市场，降低成本提高效益成为SaaS提供商迫切需要解决的问题。把传统软件迁移成SaaS软件和在IaaS（Infrastructure as a Service，基础架构即服务）平台上部署SaaS服务是SaaS服务运营前后首选的降低成本的方法，也是近几年云计算技术的研究热点。本文主要对单租户数据库模式向多租户数据库模式的SaaS迁移和IaaS平台上的数据库系统资源供给机制进行了研究，着重进行了以下几个方面的工作：首先，为了满足租户对数据模式性能和成本的双重需求，本文通过基础模式和元信息数据库的协作设计了混合多租户数据库模式，并通过增加数据访问代理层和统一数据库服务接口实现了该混合模式的轻量级迁移。实例系统的迁移实现证明了该方法能够让迁移后的SaaS软件数据库服务具有多租户访问、安全性、可配置性和可扩展性等特性。其次，本文提出了IaaS平台上多租户共享数据库的租户安置和调整算法SmartDeploy和SmartAdjust，通过在新租户进入时的租户安置和老租户性能报警时对租户分布位置的调整：尝试报警虚拟机上租户所有可能的调整方法，估算调整后的系统效益变化和数据调整成本，找到效益值提高最大的调整方案，从而优化SaaS提供商的系统效益。模拟对比实验结果表明，该算法对SaaS系统效益优化有良好的效果。最后，针对现有迁移研究中没有实现的租户数据个性化定制和WebServices个性化定制，本文提出了基于实例系统的实现方法，为其它SaaS软件迁移提供了借鉴实例。更多还原

【Abstract】 With the rapid development of the business model of cloud computing, more andmore software companies enter the SaaS service market. It’s urgent to address the problemof reducing cost and improving benefit for the SaaS vendors. Migrating traditionalsoftware into SaaS software and deploying SaaS services on the IaaS platform are thepreferred methods to reduce costs before and after running the SaaS services, and they arealso the most popular research topics in recent years in the field of cloud computingtechnology.This paper focuses on the SaaS transformation from a single-tenancy database modelto a multi-tenancy database model and the resources supply mechanism for databasesystem on the IaaS platform, the main content includes several aspects as follows:Firstly, to meet the tenants’ cost-saving expectation versus database performancerequirement, a mixed multi-tenancy data model is proposed by the cooperation of threeexisting basic data models and meta-information management database in the paper. Usinga database access proxy and a general data model interface, a light migration approach isdesigned and implemented to transform the new mixed data model. The case studied in thepaper shows that the approach can make the database component of SaaS applicationsmulti-tenancy, secure, scalable and customizable.Secondly, resources supply algorithms called SmartDeploy and SmartAdjust areproposed for the database system which uses the multi-tenancy shared schema data model.The algorithms adjust the resource and the current distribution of the tenants just when anew tenant comes and old tenants meet the alarm of performance. The basic means ofSmartAdjust is to find the best adjustment plan by estimating the benefit and migrationcost of every possible adjustment plans of all the tenants in the warning virtual machine.The one which increases the maximum of the benefit can optimize the benefit for the SaaSvendors. The simulation experiment shows that our algorithms can optimize the benefit ofthe system better. Finally, based on the example system, the paper proposes some methods to achievethe migration goals including the customization of tenants’ data and the customization ofWebServices which are not implemented in the existing migration works. The methodsapplied in the example are worth learning for other applications which are waiting for SaaSmigration.更多还原