An integral approach for complete migration from a relational database to MongoDB

Abstract

Today, computing has become an obligation in the lives of individuals and institutions alike. This magical sector uses and develops very rich, important, and sensitive tools and solutions, which make everyone's life easier. Computers with their tools have become indispensable and they have transformed the world into a small town. These tools revolve around data, the type, nature, and size of which continually increases up to zeta bytes of storage capacity, thus creating BigData. Considering that Relational databases present difficulties in the management of BigData, and Knowing that companies want to keep their data accumulated over decades of exercises and studies on the market, and also taking into account the cost impact if they kept both systems in terms of software, technical support and user training; therefore it is legitimate to find a reliable way to migrate their data from the relational system to a NoSQL system, which designed specifically to handle BigData. In this optic, several studies and approaches have been developed, but they present a lack or weaknesses in the treatment of the main components of the database, which we are going to deal with in our new and integrated approach to migration from the relational database system to MongoDB as a NoSQL system. In this article, we will present our contribution by developing a complete concept of our approach, starting with an introduction, which will be followed by a discussion of what other researchers have done in this direction, then we move on to a phase of analysis and modeling to develop the models and meta-models of the two systems: source and destination, also during this section we also present our analysis and modeling methodology, to present our global approach which divides its treatment into three axes, each of which processes a part of the RDBMS with a particular nature of data: data stored in tables, data carried on the structure of the RDBMS and data coming from the semantics of relational databases. The overall architecture of our approach, which we named ''TMSDRDND'', is formed by two layers: ''TSRSNLayer'' which deals with the transformation of the structure and the transfer of semantic data, and the ''MDRSNLayer'' layer, Which takes care of data migration using an ETL to be developed according to a specific concept and architecture and exploiting the results of the first layer.