The model-assisted global query system
Waiman Cheung · 1992
Today's enterprises typically employ multiple information systems which are independently developed, locally administered and different in logical or physical designs. Therefore, a fundamental challenge in enterprise information management is the integration of such systems across functional boundaries within an organization. Conventional technologies, such as syntax-based query languages and heterogeneous distributed databases (HDDBMS) are not sufficient to solve this problem. This thesis develops a new approach, the model-assisted global query system, that utilizes enterprise metadata to facilitate global query formulation and processing. A definitive model that characterizes the various classes and roles of knowledge in terms of these metadata is presented. The significance of possessing this knowledge (via a metadatabase) for the notion of intelligence and assistance is analyzed. Finally, the utilization of this new approach for major, difficult tasks in global query operations is established. Using this on-line knowledge, new methods are developed and verified using a prototype system to provide non-syntax-based global query capabilities to enterprise users. Specifically, these methods include (1) a model traversal method that allows direct articulation in terms of information models for global query formulation, (2) a knowledge processing method that detects and resolves conflicts using a rule processor, and (3) methods for global query optimization, implicit joins determination, and results integration. The model-assisted approach also resolves some interoperability issues (e.g., differences and incompatibilities among local systems) in information sharing. Traditional HDDBMS approaches handle these differences by imposing an integrated schema on the local systems. This approach, instead, shows that the same objective of resolving differences can be accomplished by utilizing the metadatabase containing data semantics, conversion rules, and other knowledge of the system. This way, not only difficult tasks such as schemata integration at both design time and run-time can be largely avoided, but also a higher degree of flexibility of system evolution can be achieved. The latter is an important goal of open system architecture. In sum, the Model-assisted Global Query System approach contributes to the problem of global queries in heterogeneous, distributed environments. The direct method in its own right also supports a high-level user interface (e.g., decision-support query formulation) without requiring technical details. This approach can be further generalized to facilitate other kinds of interoperability tasks in general software environments including distributed computing concurrent application systems, and other paradigms that involve interpretations and translations between the globally formulated structures and their local implementations.