Transforming UML Class Diagram into Cassandra Data Model with Annotations

The coming of big data era seriously challenges many traditional database techniques and it also explains the popularity of NoSQL databases, Cassandra, for example. Storing data in these NoSQL databases is a prerequisite for applying some popular large-scale data processing frameworks such as Hadoop or Spark. In this paper, we propose a model transformation approach that transforms class diagrams into Cassandra database schemata, which make it possible to obtain larger storage capacity and large-scale data processing ability without painful database schema redesigning. To achieve this goal, meta-models of the source model and target model are built by simplifying the definition of the UML class diagram and studying the Cassandra data model respectively. An annotation system is designed to improve the transformation, i.e., a more appropriate schema for Cassandra and expected runtime performance are achieved by adding extra information about query patterns to the transformation phase. This transformation is implemented with the ATL Transformation Language (ATL).