Performance tuning of Unix systems using knowledge-based technology

Abstract

Use of knowledge-based systems have been shown to be of relevance in a number of application domains such as in medicine (Mycin) and mineral exploration (PROSPECTOR). However, their use in the area of performance evaluation and tuning of Unix-based systems appears to be relatively unexplored. In this paper, a number of areas in which tuning efforts can be directed are identified and a knowledge-based system was developed based on the associated knowledge elicited from experts in the respective areas. An expert system development methodology was abstracted from the literature and employed in this work. The design considerations of the system are also described. Finally, possible future work and enhancements on the system are discussed.