File design of small real-time computer system

Abstract

Data loss is always a basic concern in designing a file system. Usually in large computer systems, many preventive and recovery procedures are prepared. Sometimes they may require the intervention of a trained operator. But in small computer systems where casual operators are most likely users, nothing can be expected from them in recovery. Here a file system, which assumes “Assurance of user data integrity and automatic recovery against any unexpected power failure” as a design point, which comes from the characteristics of the working environment of small computer systems, has been designed. The concepts of “System log”, “Deferred update”, and “Checkpoint” were introduced into this file system. The methods to realize the design concepts are also presented.