Multi-model Database Design and Query Processing in NewSQL DBMSs

NewSQL DBMSs are hybrid systems that combine the advantages of both SQL DBMSs and NoSQL DBMSs. This paper proposes a method for designing a database to be implemented on a NewSQL DBMS. The objectives of this method are identifying and defining heterogeneous collections of values that adhere to different data models, which are essential for an enterprise's operations, with the goal of storing and managing them within only one database. This method is based on a Relational Data Store and the Nested Relational Model. It allows the designer to use the Data Store as a guarantor of integrity and to optimize it for hybrid workloads (transactional and analytical), performance, scalability, and continuous data availability. As for the nested relational model, it is used by the designer to: (1) clarify their choices regarding storage models that can enable fast access to data about complex real-world entities; (2) specify access paths that can meet user needs. The main interest and the originality of this methodological approach are that this enables us to use the Nested Relational Model as a Pivot Model to: (1) automatically generate the global external schemas of the NoSQL virtual databases, allowing users to view and manipulate the Data Store as if it were a NoSQL database (object-relational, XML, JSON, or graph-oriented), and (2) unify the processing of cross-model SQL queries through an innovative and efficient approach. This method consistently integrates, through five levels of abstraction, the design process of the relational Data Store and that of the virtual databases. The research method used consisted of: (1) defining the objectives of this approach, (2) identifying the required levels of abstraction in light of the targeted objectives, (3) determining, for each level of abstraction, its specific objectives as well as the role to be played by the designer, a design support tool, and the DBMS, and (4) applying this approach to a typical example reflecting the most common needs, in order to facilitate the understanding of its contributions and relevance with respect to the intended objectives.