The large data base file structure dilemma.

This paper first presents a brief tutorial on the principal random file organization methods for handling two major applications--Transaction oriented systems and information storage and retrieval systems. It then addresses a particular large data base dilemma, not satisfactorily resolved by any of these methods, and which is currently under active investigation. Two approaches to a solution are described. One is called the hybrid inverted list; the other is based upon an old technique called super-imposed coding. The former has been implemented and has recently been installed in an operational system. Some statistics related to file characteristics in this application are provided, but operational cost and performance statistics are not yet available.