An Algorithm for Mapping the Relational Databases to Mongodb – a Case Study

The paper presents an original algorithm for automatic mapping of relational databases to MongoDB NoSQL databases. The algorithm uses the metadata stored in the relational system tables. It takes into consideration the concepts from Entity-Relationship (ER) model: entity type represented by a relation in the Relational Model (RM), 1:1 and 1:M relationship type represented with Foreign Keys (FK) in the RM and N:M relationship type represented in RM with a join table that contains the Primary Keys (PK) from the original tables, each representing a FK and two 1:M relationships between the original tables and the join table. Such an algorithm is very useful because in practice there are already implemented relational databases with a big number of records. It was proofed that for some of them the solution of implementation using document databases like MongoDB is a better one, from the management point of view, but also regarding the query time response. A significant amount of time and human effort is spared this way, which would have otherwise been needed to create and populate the NoSQL database from scratch Such an example is a hospital database that manages patient sheets that have different structure at different sections. In this stage, the algorithm was tested on a MySQL database containing 10-15 tables with many relationships between them and approximately 100 records/ table.