Font Size: a A A

Research On Shared Storage Technology For SaaS Multi-Tenant Data Isolation

Posted on:2019-05-01Degree:MasterType:Thesis
Country:ChinaCandidate:B PengFull Text:PDF
GTID:2428330596966401Subject:Software engineering
Abstract/Summary:PDF Full Text Request
With the rapid development of the Internet and cloud computing technologies,more and more Web systems use the SaaS(Software as a Service)platform.As the scale of tenants gradually increases,the amount of data needing to be stored on the SaaS platform will become very large.If we use the traditional branch store table storage mode,it will bring huge storage cost.The current research trend is to use a shared database shared data mode of storage,which has the advantage of being able to guarantee the isolation of data between different tenants,but also allows tenants to maximize the sharing of data architecture,and saving storage costs.This research focuses on studying multi-tenant data sharing storage model,mainly to solve the following problems:(1)Enable different tenants to share data sharing patterns in the database while protecting the isolation between tenants;(2)Improve the customization capabilities of shared storage models,mainly aims at the ability to meet tenants' new customization requirements;(3)The existing shared storage model lacks the consideration of the two functions of schema mapping and SQL rewriting,and only analyzes the theoretical feasibility.Therefore,this research analyzes the existing deficiencies of the existing shared storage model,proposes two improved shared storage models,and provides API support for the mode mapping layer.This thesis presents the main research work as follows:(1)When the existing shared storage model meets the needs of the tenants for customized transformation,the solution adopted is to reserve columns and perform data migration,which will greatly affect the performance of the database and the use of other tenants.This thesis studies the mode mapping mechanism and designs two shared storage models: wide table + chunk table,wide table + pivot table,and the strategy of constructing multiple wide tables,chunk tables,and pivot tables.(2)Combined with the hierarchical structure of shared storage,the working principles of the two types of shared storage models,wide table + chunk table,wide table + pivot table,are analyzed;and through analysis of the mode mapping process and relational algebra operations respectively,the equivalence between shared storage model and traditional relational model is established.(3)For the schema mapping mechanism and logical SQL rewriting mechanism in the schema mapping layer,an API that supports two major functions of field mapping and SQL rewriting is designed and implemented,which provides effective guarantee for shared storage and logical isolation of multi-tenant data.(4)The experimental test and result analysis of storage space utilization and query efficiency for two types of shared storage models: wide table + chunk table,wide table + pivot table,are carried out.The function of field mapping and SQL rewriting in two shared storage models are respectively verified.
Keywords/Search Tags:SaaS, Multi-tenant, Shared storage model, Pattern mapping
PDF Full Text Request
Related items