Towards customizable, flexible storage structures for complex objects

Abstract

During the last years several new data models have been developed which directly support complex objects, especially their structure definition and manipulation. In most of these approaches the internal storage structures are directly derived from the logical structure of the complex objects following a fixed System inherent mapping strategy. The adequacy of a certain mapping strategy, however, is strongly dependent on the access patterns of the related applications. It should, therefore, be possible to define the internal representation (= physical schema) of complex objects irrespective of their logical structure (= external Schema). In this paper we discuss appropriate base constructs for implementing complex objects and describe how to define alternative internal storage structures related to these base constructs by just specifying appropriate parameters in a respective data definition language. In addition, we also discuss how to control the clustering of these objects. Altogether, we achieve a flexibility which allows to define for a given logical complex object structure almost every storage and clustering structure discussed in the literature. Further-more, many other variants and mixtures of storage structures not discussed yet can be defined.