Research On Self Evolution Of Database Schema Based On Multitenant Application

Multi-tenant SaaS (Software as a Service) application for its on-demand, pay, no maintenance and other characteristics, the era of cloud computing is becoming the main software service model. The so-called multi-tenant SaaS application is single-instance multi-tenancy. In other words, an application instance for deploying a lease with multiple tenants, tenants can be customized to meet their own personalized applications, custom mainly from the application interface, data model, service procedures, and several other level to start, where the data model is a multi-tenant application customization customization of the core, the core elements of the business custom.However, with the expansion of the tenants, customized various tenants a large number of data patterns in the same pattern of public information, which is further tenants custom modifications, and new tenants inconvenience customization process, because all must be based on original custom the basis of common data model began to be customized, custom data models new tenants increasingly heavy burden, and even a large number of tenants customization is repeated, seriously affecting the tenant’s data model customized experience, increasing the complexity of the customization process. In response to these problems, in order to improve the application of large-scale multi-tenant tenant customization efficiency. Articles about multi-tenant application data patterns from the evolution of in-depth research. The so-called multi-tenant data pattern from the evolution of the application is the multi-tenant applications in a shared data storage, the base of the common shared data pattern can customize various tenants with information to add, to automatically analyze and extract the common information between the multi-tenant, the common data model based adaptive evolution, to reduce future tenants customization steps to improve the efficiency of multi-tenant applications customization. Multi-tenant application storage architecture usually use shared database shared data schema. this architecture will store isomorphic tenant’s data in the basic data table and store heterogeneous data in extended data table. Data access efficiency in extended data table is relatively low, so that extended data table scale directly affects the overall system performance. The early stages of system operation, the basic structure of the table data to meet the basic needs of the tenants, extended data table is relatively small store data. With the changing needs of the tenants, the tenant custom extension field increases, the scale of the extended data table is also growing, so the system performance will be seriously affected.To this end, the article in the shared tables and shared database instance architecture is proposed based on multi-tenant application evolution method since, first, on the basis of a common data model continuous self-evolution, it has been basically meet the needs of the majority of tenants, new tenants to reduce the workload of data patterns customization. Second is the basic data sheet and extended data field in a table for self-evolution of the division, it is necessary to guarantee the basic data sheet can not be too sparse, but also reduce the maximum possible expansion of the scale of the data in the data table.By developing experimental system to verify the basic public evolution effectively improve the tenant from custom data model experience, shared data model proved both to improve the self-evolution of data storage utilization, but also improve the overall multi-tenant application data access performance.