Research On Multi-tenant Data Storage Mode Of SaaS Application Based On Universal Table

Software as a Service(SaaS,Software as a Service)is a new type of software application mode.As one of the main service forms of cloud computing,software services are provided online through the Internet and users do not need to perform local installation and deployment.Multi-tenant technology is the core technology of SaaS applications.Tenants realize the reuse of hardware and software resources through “single instance multi-tenancy” to maximize resource utilization and minimize cost.Under the same application instance,different tenants have different data schemas,so how to access data with different schemas but belong to the same application becomes the primary issue for SaaS multi-tenant data storage.The existing multi-tenant database design scheme has Universal table,pivot table,private table,and extended table,and has its own advantages,disadvantages,and application scenarios.The pattern mapping technology based on a relational database cannot take into account system performance and storage.Overhead and scalability requirements.Therefore,this paper presents an improved multi-Universal SaaS application multi-tenant data storage management mechanism,describes its data definition,mapping mechanism,query rewrite process,and indexing mechanism in the multi-Universal table mode,which can be effective while saving storage costs.Improve storage space utilization and query performance.The multi-universal table data storage mode shares and stores data of different tenants,resulting in a large number of null values due to the sparse data of the columns.At the same time,the schema mapping is complicated and query rewriting is difficult.This paper improves multi-universal table data storage mode,replaces the previous single wide table with multiple Universal tables with different number of columns,and stores data in the appropriate wide table according to the characteristics of tenant data and individual requirements,effectively reducing the number of columns.The number of nulls improves theutilization of data storage space.The data structure and pattern mapping mechanism of the multi-universal table storage mode are defined in detail,and the query rewriting process combining dynamic SQL rewriting and view definition SQL rewriting in data access is realized,and the complexity of the logical SQL statement by the tenant request is differentiated.Flexible selection of query rewriting methods reduces query rewrite difficulty and improves query efficiency.Based on the customization of SaaS applications by tenants,business data of different actual data types can be shared and stored in the same data column.Therefore,the logical view of the tenant data table is different from the physical storage structure,and column-based relational data indexing cannot be applied to multi-universal tables.Data storage mode.This paper proposes an improved Indexes Pivot Table indexing model.According to the tenant customization requirements,each tenant business data is formed into index metadata,and metadata is used to drive customization and isolation of different tenant index data.At the same time,an index maintenance strategy is given to ensure that the tenant index data is updated synchronously with the business data.Experiments show that under a single data node,quick access to data can be achieved through two-level indexing,namely tenant logical indexing and relational database physical indexing.Finally,the rationality of the multi-universal table mode storage mechanism and the effectiveness of the index mechanism are verified through experiments.The results show that the query efficiency of the tenant index operation is significantly better than that of the non-index environment,which improves the overall utilization of storage space and improves the data query efficiency.