Fragmented Database Management Systems

Greg Pytel

A concept of Fragmented Database Management System (FDBMS) is primarily based on Distributed Relational Database Management System (DRDBMS), augmented by elements of Object Oriented DBM's. One of the main objectives of DRDBMS is to set up a theoretical framework that allows analysis of physical information (i.e. information that we need to store and manipulate) in order to organise its management - this includes optimisation of system performance, information availability, etc. through data distribution. The logic of this process can be compared to "top-down" approach used in compilers methodology (i.e. from global view of the information - global or world schema - towards the consistent distribution into individual fragments - local schema.) The DRDBMS approach is practical primarily when it is used to implement new and single information system. The reverse situation is when there already exists number of information systems with their own schemas (semantics of attributes, functional dependencies, etc.). Creating in such environment a single, integrated information system is a partial change of paradigm from "top-down" approach into "bottom-up". This change of "direction", however, brings to light very interesting issues. Fragmented Database Management System is an attempt to put a systematic view on them.