page 1  (20 pages)
2to next section

Object-Oriented DBMS and Beyond

Klaus R. Dittrich and Andreas Geppert

Department of Computer Science, University of Zurich

Abstract. Over the past 10+ years, object-oriented database systems have gone a long way from research prototypes to commercial products to real-life mission-critical applications. Currently, we also witness the extension of relational systems with salient object features, resulting in so-called object-relational DBMS.
In this paper, we introduce and review the salient features of both approaches, discuss their merits and shortcomings, and for which kinds of applications they are best suited. We also elaborate on further necessary improvements of the current state of the art. Furthermore, we will speculate about several upcoming areas of database research in a broad sense (like global information systems, workflow management, component technology) where object-orientation and object-relational and object-oriented database systems in particular might (and should!) play a leading role.

1 Introduction

Database management systems have a long tradition of research and successful deployment in computing practice. The classical" systems (hierarchical, network, and relational) are established and viable platforms for data management in a large variety of application areas. With the advent of object-orientation and new requirements imposed by advanced application domains, research has proposed object-oriented database systems beginning in the mid-80s. Since the beginning of the 90s, such systems have been commercially available. After a period of discussion (some may even say religious warfare") about whether object-oriented DBMSs are actually needed or whether relational systems can meet these new requirements just as well, relational vendors have recently reacted by incorporating object-oriented features into their systems. These extensions are also considered in the forthcoming SQL3-standard.

In this paper, we review object-oriented concepts in database systems and hypothesize about future trends in database systems with respect to objectorientation. Particularly, we argue that database technology will play an increasingly important role in new and emerging areas, such as workflow management and information integration. These areas definitely need database services, and object-oriented database technology is well-suited to meet their requirements such as modeling power and seamless integration. Nevertheless, database functionality as required by these areas can no longer be sufficiently provided by