Migration of Relational Database to NoSQL Document-Oriented Database

Migration of Relational Database to NoSQL Document-Oriented Database

연구 분야: Databases

학회: 2023 IEEE International Conference on Data and Software Engineering (ICoDSE)

With the increasing popularity of NoSQL document-oriented databases, there is a need to migrate from relational databases to NoSQL document-oriented databases. One of the main advantages of document-oriented databases is their ability to perform embedding and referencing. Stanescu et al. conducted research on a migration algorithm from MySQL relational databases to MongoDB NoSQL databases, utilizing relational database metadata. However, the algorithm developed only relies on whether tables have foreign keys and/or are referenced by other tables, which can lead to suboptimal mapping of the migration outcome model. Considering that a database migration is only performed on a database that has been or is currently in use, the workload information concept used in the research by Lima and Mello can be adapted to apply a better data mapping model. This paper aims to develop a migration algorithm by leveraging workload information to determine the use of embedding and referencing. The algorithm consists of two parts: a schema conversion algorithm that produces a schema for the NoSQL document-oriented database, and a data mapping algorithm that generates JSON files. Additionally, the developed algorithm incorporates a table evaluation sequence to complement the existing migration algorithm. The performance evaluation shows that, in general, the modified algorithm provides better performance than Stanescu et al.'s algorithm. However, this is not the case for scenarios involving high-frequency access to read embedded documents.