Research And Implementation Of Unified Query Model About Relation Database And Non-Relation Database

For the new requirement of big data management,there are many NoSQL database for specific application.Because SQL and NoSQL storage solutions have their own advantages,SQL and NoSQL mixed database solutions emerge as the times require.The application data stored in relational and non-relational database involves the management of multiple data source.Wide attention and research of enterprises and scholars on how to use hybrid database more conveniently and effectively is also more and more,much of the present database products and their adaption schema,are either commercial,or not transparent,or for a specific SQL and NoSQL database,can not provide a universal access adaptation scheme for hybrid database.Therefore,this paper designed a universal adapter to access SQL and NoSQL solutions,research and implement a unified framework to access SQL and NoSQL database,the framework is independent of the specific SQL and NoSQL databases.By the unified query model,we can operate the two database as a single one.Firstly,paper describes the theoretical framework of the model.Specifically,this paper refers to RDF concepts,such as triple,storage solution and query language,by translating the NoSQL data into triples,then merges these triples in a SQL database as a virtual relationship,treats the NoSQL database as a SQL relational table.Through the SQL statements translated from complex query language to manipulate the two data sources.Then the original NoSQL data is built by a series of self-joins of these triples.Secondly,according to the theoretical framework,a specific implementation based on PostgreSQL and MongoDB is given,and the paper provides a detailed description of the triple conversion and the translation of complex query language to SQL statement.The framework receives a user query,finally outputs an integration result from different data sources.Finally,based on the realization of the basic functions of the framework,a series of query optimization strategy is proposed,by reducing the transmission size of data from the NoSQL database and the times of self-joins of triple table to improve query performance of the query model.Experimental results show that a general unified query theoretical framework is feasible and prototype system developed follow the framework achieve to access SQL and NoSQL database uniformly by SQL-statement.Also query optimization strategies of the model can significantly improve the query performance of the prototype.It has certain practical value and engineering significance.