Active database behaviour : the REFLEX approach
Modern day and new generation applications have more demanding requirements than traditional database management systems (DBMS) are able to support. Two of these requirements, timely responses to the change of database state and application domain knowledge stored within the database, are embodied within active database technology. Currently, there are a number of research prototype active database systems throughout the world. In order for an organisation to use any such prototype system, it may have to forsake existing products and resources and embark on substantial reinvestment in the new database products and associated resources and retraining costs. This approach would clearly be unfavourable as it is expensive both in terms of time and money. A more suitable approach would be to allow active behaviour to be added onto their existing systems. This scenario is addressed within this research. It investigates how best active behaviour can be augmented to existing DBMSs, so as to preserve the investments in an organisation's resources, by examining the following issues, (i.) what form the knowledge model should take, (ii.) should rules and events be modelled as first class objects, (iii.) how will the triggering events be specified, (iv.) how the user will interact with the system. Various design decisions were taken, which were investigated by implementation of a series of working prototypes, on the ONTOS DBMS platform. The resultant REFLEX model was successfully ported and adapted onto a second POET platform. The porting process uncovered some interesting issues regarding preconceived ideas about the portability of open systems.