摘要：Relational database has been the de-facto database choice in most IT applications. In thelast decade there has been increasing demand for applications that have to deal with massive andun-normalized data. To satisfy the demand, there is a big shift to use more relaxed databases in theform of NoSQL databases. Alongside with this shift, there is a need to have a structuredmethodology to transform existing data in relational database (RDB) to NoSQL database. Thetransformation from RDB to NoSQL database has become more challenging because there is nocurrent standard on NoSQL database. The aim of this paper is to propose transformation rules ofRDB Schema to various NoSQL database schema, namely document-based, column-based andgraph-based databases. The rules are applied based on the type of relationships that can appear indata within a database. As a proof of concept, we apply the rules into a case study using threeNoSQL databases, namely MongoDB, Cassandra, and Neo4j. A set of queries is run in thesedatabases to demonstrate the correctness of the transformation results. In addition, thecompleteness of our transformation rules are compared against existing work.
