Database design practices for inverted files

Abstract

Database research literature has proposed many procedures, both manual and automated, for database design; selection of secondary indexes for inverted file type database management systems (DBMS) has been repeatedly addressed. The empirical study reported here indicates that practical inverted file design has been relatively unaffected by this research. This paper characterizes the actual database design process used at inverted file DBMS installations along such dimension as: types of secondary keys constructed, the individuals who make index design decisions, the decisions that are changed (and when) after the initial database implementation, the factors that are considered in indexing decisions, and the literature which is used in the process. The study shows that key selection (as one example of a design decision) is addressed by ad hoc procedures and well conceived procedures are not used. Further, the results indicate that database design is dominated by users and systems analysts, indexes are frequently changed and a wide range of database performance and convenience factors are influential in practice. The paper concludes with some recommendations for database design support tools.