Differential Files: Applications and Analysis.

Abstract

The representation of a collection of data in terms of its differences from a preestablished point of reference is a basic storage compaction technique with wide applicability. This dissertation studies a differential file architecture which has been suggested as an efficient means of storing large and volatile databases. A comprehensive analysis of this scheme is provided, and key decision variables in the design of an efficient differential file architecture are identified. An analytic model for the differential file operation is developed. Solution procedures, based on st and ard numerical search techniques and heuristics, are provided which can quickly generate (near) optimal differential file design. Applying the results to realistic problems, the potential for a modest reduction in operating cost, and significant improvements in database recovery and access times are demonstrated.