A New Approach for an Efficient Relational to Document-Oriented Database Migration

Businesses increasingly encounter scenarios demanding adaptation to unstructured or semi-structured data, scalability improvements, and agile development methodologies, prompting a compelling need for transition to document-oriented databases. This transition is an ongoing process for many organizations, but it comes with various kinds of challenges such as complexity of data migration procedures like Extract, Transform and Load (ETL) processes, maintaining data integrity and schema transformation. In this paper, we present a new Model-Driven (MD) approach for Relational to Document-Oriented Database Migration embedded into our RelToDocMigrator development environment. We have developed a domain-specific language, named MigDSL, used to define database migration model, as a part of the developed RelToDocMigrator tool. MigDSL allows developers to work at a higher level of abstraction, with concepts and structures that are closer to the end-users and their perception power. Our approach provides a high degree of flexibility and customization, allowing developers to tailor a migration process to specific needs and write a lower amount of code compared to general-purpose programing languages (GPLs). By developing migration model, users automatically get document-oriented schema for data validation before inserting extracted data into document-oriented databases. Based on the created database migration model, the executable programming code is automatically generated using the code generator. The code generator produces Python programming code and can be extended to produce migration code in various languages. The generated migration code may be restructured by developers when needed and executed to create an improved database schema.