Vertical partitioning is a design technique for reducing the number of disk accesses to execute a given set of queries by minimizing the number of irrelevant instance variables accessed. This is accomplished by grouping the frequently accessed instance variables as vertical class fragments. The complexity of object-oriented database models due to subclass hierarchy and class composition hierarchy complicates the definition and representation of vertical partitioning of the classes, which makes the problem of vertical partitioning in OODBs very challenging. In this paper, we develop a comprehensive analytical cost model for processing of queries on vertically partitioned OODB classes. A set of analytical evaluation results is presented to show the effect of vertical partitioning, and to study the trade-off between the projection ratio versus selectivity factor vis-a-vis sequential versus index access. Furthermore, an empirical experimental prototype supporting vertical class partitioning has been implemented on a commercial OODB tool kit to validate our analytical cost model.

In this paper, we introduce structural join index hierarchy (SJIH) mechanisms that mimic the class composition hierarchy of the complex objects to provide direct access to complex objects and/or their component objects.SJIH extends the work on join index hierarchies by providing efficient and flexible access to complex objects, thus unifying various previous indexing methods proposed for object oriented database (OODB).A cost model for processing the queries and maintaining the indices is developed.A heuristic algorithm is designed to select appropriate indices to efficiently process a given set of queries.We show that SJIH indexing mechanisms facilitate efficient retrieval of complex objects.Given a set of queries and a limited index storage space, the heuristic algorithm facilitates fast selection of near-optimal set of indices for efficiently executing the queries.

Conventional object-oriented database (OODB) systems follow a class-based approach to model real world entities and their relationships, which imposes several difficulties when more advanced "dynamic functions" are needed by the applications. One of such dynamic functions is conceptual clustering — the ability to dynamically group a set of existing objects together as a higher-level abstract object (called a "cluster"). The conceptual clustering model (CCM) that we have defined is an extension to current object-oriented models, which facilitates the modeling of dynamic, evolving, and/or ad hoc object clusters through a well-devised clustering mechanism that incorporates "roles". This paper presents the advanced features of CCM, along with its associated query language facilities devised for supporting general definition and access of clusters (and roles). An experimental prototype of the CCM has been constructed on top of a persistent object storage manager, running on a Sun4 workstation.