多租户数据管理关键技术研究

随着云计算的日益普及和大数据应用的迅速发展，越来越多的企业与个人选择“数据库即服务”方式来管理数据。这种外包化的数据管理模式在云端形成了资源共享的多租户环境。如何在多租户环境中有效管理海量数据和处理动态负载不仅是服务提供商需要面对的问题，也是数据库界面临的挑战，因为多租户环境的独特性使得传统数据管理技术难以直接有效地移植到多租户环境中，因而必须研究专门的多租户数据管理技术。本项目旨在通过发掘和利用多租户平台的特点和租户负载特性，攻克多租户环境下的租户负载分析与建模及预测、数据组织与存储及索引、数据缓存、数据查询处理、资源优化配置与负载均衡等关键技术问题，实现对多租户负载的低成本、高效率处理，为多租户数据的有效管理提供良好的技术支撑。

With the increasing popularity of cloud computing and the rapid deployment of big data applications, more and more businesses and individuals turn to Database as a Service (DBaaS) platforms to manage their data, and such out-sourced data management applications lead to multi-tenant environments in clouds, where system resources are shared by all tenants. How to effectively perform massive multi-tenant data management and efficiently process the dynamic workloads of tenants is not only a crucial problem that service providers need to deal with, but also a huge challenge that the database academia has to face, as the unique characteristics of multi-tenancy makes traditional database techniques not effectively applicable in multi-tenant environments, so data management techniques specifically for multi-tenant environments are required. This proposed project aims to tackle the key techniques for multi-tenant data management, based on a comprehensive analysis and deep understanding of the characteristics of multi-tenant applications and tenant workloads. Research issues include: analysis, models and prediction methods of tenant workloads, tenant data organization, storage and indexing, data caching mechanisms for mixed tenant workloads, query processing algorithms, resource configuration and loading balancing. This study will incubate innovative techniques for efficiently processing multi-tenant workloads in a cost-effective way, and effectively supporting multi-tenant data management, drive the development of mult-tenant database applications.