This paper describes the cost-performance of an information retrieval system where inverted list processing is off-loaded from a minicomputer to the user’s 16-bit microcomputer via a direct asynchronous connection. Experimental data are used to show that response time for the distributed system is better than a comparison centralized system when the minicomputer is heavily loaded and when transmission rates to the microcomputer are greater than 2400 baud.
The analysis and approach taken in this paper are very simplistic. The experimental database used contained only 936 records and the reference lists transferred to the microcomputer were never longer than several hundred entries. No consideration is given to the cost of the microcomputer when comparing the cost of queries using the distributed system. The authors mention that they were able to study the software engineering problems involved in implementing their system but never describe these.