A decision support system for database evolution using data model independent architecture

Abstract

The existing corpus of scientific knowledge in the field of database systems (DBS) has been concerned mostly with such problems as database technologies and system development methodologies. Relatively few efforts have been devoted to the problem of adapting an ongoing DBS in a systematic fashion. Notwithstanding the lack of sufficient prior knowledge, this adaptation problem is critical in DBS management, since a DBS should really be conceived as an evolving structure, rather than a stable one-shot phenomenon. Toward this end, this paper proposes a reliability-based decision support framework for evolving a DBS systematically. Both user satisfaction with the DBS and the usage pattern of it are monitored on a real-time basis and used for controlling it adaptively. Chance-constrained models are proposed to characterize suitable decision rules for DBS evolution decisions ranging from file reorganization to DBS restructing. A four-schema architecture is also proposed for achieving data model independence in a DBS, whereby facilitating the control and evolution of DBS.