Evaluation and optimization
Elias Berelian, Keki B. Irani · Very Large Data Bases · 1977
A mathematical methodology for data definition design is introduced. The objective of this design is to produce a data structure for a paged data base that conforms to the Data Base Task Group specifications. The design methodology takes, as input, a high level description of the data base. It then produces a data structure which is optimal over a certain class of DBTG data structures that describe the data base, in the sense that the expected page fault rate is minimized for a given set of applications and a given storage bound. The optimization takes into account such factors as (data base) set implementation techniques and applications' security requirements. The algorithm used to solve the optimization problem uses a dynamic programming approach, based on the idea of step-wise generation of undominated solutions.