SaaS平台多租户数据管理及逻辑存储模型的研究
发布时间:2018-07-25 07:16
【摘要】:SaaS(Software as a Service)平台应用的业务领域千差万别,对应的数据模式也各不相同,对于每一个SaaS应用,虽然不同租户其业务不一样,但同一个应用下数据模式相似。这就要求SaaS平台在保证性能和应用高效运转的前提下支持数据存储的弹性扩展。 首先,本文总结出多租户云数据管理需要考虑的多租户数据的特性,结合多租户数据存储的特性,提出有效支持多租户个性化的SaaS平台多租户数据管理方案。 其次,通过对现有的通用表、透视表等模式进行研究比较,分析其定制能力及导致性能下降的关键因素,对透视表存储模型进行改进,提高定制能力,从而形成新的多租户共享数据存储模型。基于新的数据存储模型,实现租户逻辑数据与物理存储模式之间的模式映射和查询转换机制,达到定制能力与系统性能平衡的目的。通过实验,测试数据库中表的数量对数据库吞吐率及缓冲池命中率的影响,并对提出的数据存储模型与透视表模型的查询转换性能进行对比测试,验证了本文提出的数据存储模型的有效性。 最后,在实验室承担的华为创新项目“云计算及其支撑网络项目”的业务层子系统的基础上,以多租户个性化的数据管理方案及透视表改进模型为理论基础,设计和实现SaaS平台多租户数据管理模块。
[Abstract]:The business domain of SaaS (Software as a Service) platform application is very different, and the corresponding data pattern is different. For each SaaS application, although different tenants have different business, the data pattern is similar under the same application. This requires the SaaS platform to support the elastic expansion of data storage while ensuring the performance and efficient operation of the application. Firstly, this paper summarizes the characteristics of multi-tenant data management in multi-tenant cloud data management, and proposes an effective multi-tenant data management scheme for multi-tenant SaaS platform based on multi-tenant data storage. Secondly, through the research and comparison of the existing general table, perspective table and other modes, the customization ability and the key factors leading to the deterioration of the performance are analyzed, and the perspective table storage model is improved to improve the customization capability. Thus a new multi-tenant shared data storage model is formed. Based on the new data storage model, the schema mapping and query conversion mechanism between the tenant logical data and the physical storage mode is realized, which achieves the purpose of balancing the customization capability and the system performance. Through experiments, the effects of the number of tables in the database on the throughput and buffer pool hit rate of the database are tested, and the query conversion performance of the proposed data storage model and the PivotTable model is compared and tested. The validity of the proposed data storage model is verified. Finally, on the basis of the business layer subsystem of Huawei innovation project "Cloud Computing and its supporting Network Project", the paper takes the multi-tenant personalized data management scheme and the PivotTable improvement model as the theoretical basis. Design and implement multi-tenant data management module of SaaS platform.
【学位授予单位】:西北大学
【学位级别】:硕士
【学位授予年份】:2013
【分类号】:TP333;TP393.09
本文编号:2143047
[Abstract]:The business domain of SaaS (Software as a Service) platform application is very different, and the corresponding data pattern is different. For each SaaS application, although different tenants have different business, the data pattern is similar under the same application. This requires the SaaS platform to support the elastic expansion of data storage while ensuring the performance and efficient operation of the application. Firstly, this paper summarizes the characteristics of multi-tenant data management in multi-tenant cloud data management, and proposes an effective multi-tenant data management scheme for multi-tenant SaaS platform based on multi-tenant data storage. Secondly, through the research and comparison of the existing general table, perspective table and other modes, the customization ability and the key factors leading to the deterioration of the performance are analyzed, and the perspective table storage model is improved to improve the customization capability. Thus a new multi-tenant shared data storage model is formed. Based on the new data storage model, the schema mapping and query conversion mechanism between the tenant logical data and the physical storage mode is realized, which achieves the purpose of balancing the customization capability and the system performance. Through experiments, the effects of the number of tables in the database on the throughput and buffer pool hit rate of the database are tested, and the query conversion performance of the proposed data storage model and the PivotTable model is compared and tested. The validity of the proposed data storage model is verified. Finally, on the basis of the business layer subsystem of Huawei innovation project "Cloud Computing and its supporting Network Project", the paper takes the multi-tenant personalized data management scheme and the PivotTable improvement model as the theoretical basis. Design and implement multi-tenant data management module of SaaS platform.
【学位授予单位】:西北大学
【学位级别】:硕士
【学位授予年份】:2013
【分类号】:TP333;TP393.09
【参考文献】
相关期刊论文 前4条
1 何剑虹;白晓颖;李润玲;崔智社;;基于SLA的面向服务的基础设施[J];电讯技术;2011年09期
2 倪萍;廖建新;朱晓民;万里;;一种基于QoS的QoE到SLA映射方法[J];电子与信息学报;2010年06期
3 昌中作;徐悦;戴钢;;基于SaaS模式公共服务平台多用户数据结构的研究[J];计算机系统应用;2008年02期
4 殷康;;云计算概念、模型和关键技术[J];中兴通讯技术;2010年04期
相关博士学位论文 前1条
1 孔兰菊;SaaS应用交付平台中多租户云数据管理关键技术研究[D];山东大学;2011年
,本文编号:2143047
本文链接:https://www.wllwen.com/kejilunwen/jisuanjikexuelunwen/2143047.html