Extensibility in the Starburst database system

The goal of the Starburst project is to examine how traditional relational database systems must be adapted for new applications and technologies. We believe that extensibility is an important requirement for future database systems, and have identified some of the problems of building an extensible relational database system. We focus on five potential areas for extensibility: external data storage, storage management, access methods, abstract data types, and complex objects. Our approach emphasizes extensions written by knowledgeable programmers that are invoked as needed from within the primary database system. We show by an example how this method could be used to implement support for complex objects. Although many questions remain unanswered, extensible relational database systems are a promising area of research.