Steffen Skatulla

"Storage, Indexing and Processing of Complex Objects in ORDBMS"

Modern relational DBMS use more and more object-relational features to manage complex objects with nested structures and collection-valued attributes, thus evolving towards object-relational database management systems. But with respect to the nested complex objects, ORDBMS in general offer far too little flexibility to adapt storage structures, indexing and query execution to particular workloads. And almost all systems mix up logical design and physical layout specifications.

With the development of our Physical Representation Definition Language (PRDL) we promote the idea of an intermediate layer separating logical and physical design. PRDL allows to adapt the storage structures and the indexing of nested complex objects to particular workloads. In completion to the examination of storage and indexing structures we studied query execution issues with respect to complex objects including plan operators, algorithms, cost models and optimization.

Additionally, we present simulation and measurement results that show the performance benefit achievable adapting storage structures, indexing and query execution algorithms to different workload profiles on complex objects. The presented work is part of the research project "Object-Relational Database Features and Extensions: Model and Physical Aspects" of the Jena Database Group. The research project is supported by an IBM Shared University Research Grant.