On conceptual design of active databases
Asterio Kiyoshi Tanaka · SMARTech Repository (Georgia Institute of Technology) · 1992
An active database management system (DBMS) is a system with full database functionality plus the additional capabilities of monitoring the state of the database and executing some predefined actions when appropriate events are detected. Although this is a well-established research area, and a few commercial relational DBMSs already support some active capabilities, no support for conceptual modeling and design of active database behavior is currently provided to take advantage of the new capabilities. The current database design methodology forces the user to defer major modeling decisions concerning the active behavior of the database to late stages of the design process, where the semantics of the real-world situations are obscured by the intricacies of the implementation model. This research addresses the conceptual design problem by incorporating active database behavior into the Entity-Relationship (ER) model, in the form of events and rules. The resulting Entity-Relationship-Event-Rule (ER)$\sp2$ model provides an extended architecture of tools for assisting the database designer in the task of specifying, analyzing, and translating active behavior into executable data definition statements in specific relational DBMSs. Meta-behaviors are identified that automatically enforce invariant properties of the model, along with user-defined integrity constraints. A high-level Petri net, e/r-net, whose places are events, transitions are rules, and net inscriptions are (ER)$\sp2$ schema definitions is proposed as an analysis tool for validating the active behavior design. A prototype implementation of the extended architecture is described that takes advantage of the meta-data about the design process generated during the schema translation. The following benefits result from the extended data modeling and database design methodology: reduced database design and application development effort with the automatic generation of meta-behavior and translation of active behavior into executable DBMS language constructs; better control of the development of database applications by introducing precision in the specification of active behavior; and better quality of the overall design, with an interactive environment capturing the knowledge of application experts and automatically dealing with the DBMS constraints.