Model-driven development of OData services: An application to relational databases

Abstract

Open Data Protocol (OData) is a protocol to facilitate the publication and consumption of queryable and interoperable data-driven online services. OData is based on the use of RESTful APIs derived from a data model plus a URL-based query language to identify and filter the data described in such model. Due to its maturity and ease of use for end-users and client applications, OData has become the natural choice to publish datasets online. Still, creating OData services is a tedious and time-consuming task, since data providers should (1) represent their data models in OData format, (2) implement the business logic to transform OData requests to SQL statements (or the target storage technology of choice), and (3) de/serialize the exchanged messages conforming to the OData protocol. This paper presents a model-based approach aimed at (semi)automating all these steps. From an initial UML class diagram, we derive all the artifacts required to have an OData service up and running on top of a relational database conforming to the model definition. A prototypical implementation of the approach is provided.