A Designer For DBMS-Processable Logical Database Structures

An analytical approach to the design of logical database structures is presented. Inputs to the model consist of data item types and volume, associations among items, security constraints, and item access frequencies required to satisfy known database applications. The output is a prototype database management system (DBMS) processable schema for a hierarchical and/or network database structure. The design problem is formulated as an integer programming problem and the branch and bound method is used to generate an optimal solution for the parameters specified. The objective function to be minimized is logical record access, subject to constraints on record size and total database size. Experience with the model has shown it to be an excellent predictor of real performance, easy to use, and have relatively low computing cost for moderately-sized database design problems. Use of the model is illustrated through an example problem and its solution. Direct comparison with a previous methodology clearly establishes the superiority of the current approach.