NOSQL DATABASE LOGIC DESIGN METHODS FOR MONGODB AND NEO4J

Abstract

Modern developers of gaming mobile and Internet applications almost do not imagine themselves without the use of NoSQL databases, if they pursue the goal of creating scalable databases with high-performance and wide functionality.  When designing a database for any NoSQL system, the developer needs a clear understanding of the logic of such databases and the capabilities of the tools offered by the corresponding DBMS. However, unfortunately, they do not have unified methods of logical design of such models, as in relational databases. Thus, there is a problem of developing effective methods for the logical design of such databases that would provide the necessary performance when implementing the business logic of the corresponding applications. The subject of the research is approaches to the logical design of NoSQL document and graph databases. The goal of the work is to propose unified logical modeling methods for MongoDB and Neo4j NoSQL systems based on an experimental study of their performance. The following tasks are solved in the work: analysis of current approaches to the logical design of document and graph databases\, the development of logical design methods for them; planning and experimental study of the performance of the proposed methods on the example of models developed with their help. The following methods are used: database design methods, database performance evaluation methods, development methods are based on MongoDB 5.0.5, Neo4j 4.4.3 DBMS, Visual Studio 2022 development environment. The following results are obtained: unified logical design methods for MongoDB and Neo4j NoSQL systems are proposed; on their basis, the corresponding logical models have been developed; experimental measurements of the number of resources required working with the developed models; recommendations on the proposed methods are formed. Conclusions: The proposed modeling methods for MongoDB have their own aspects of their effective use for different types of applications. The strengths and weaknesses of both methods were identified, but a mixed method based on a combination of modeling through normalization and denormalization was recommended. Even though Neo4j lost out to MongoDB in terms of consumed resources in most experiments, both DBMS's' demonstrate good productivity, taking into account the orientation to different tasks.