Matching storage organisation to usage pattern in relational data bases

Abstract

A significant improvement in the performance of data base systems may be achieved by dynamically optimising the storage organisation and the access paths in accordance with the usage patterns. The principle of defined relations may be .employed to ensure that a given relational data base is tuned to match its usage pattern. This thesis describes some original contributions to the solution of the update problem of defined relations. Some methods of improving the response time without impairing the utilization of disk space have been investigated, and a generalised page replacement algorithm for the management of the data base work space is recommended. The arguments in this thesis are supported by examples drawn from existing relational data bases. As a whole, the thesis emphasises the benefits of organising the data base in a manner dictated by the activity of its users.