Software events tracking system

Lack of reliability is a problem that continuously plagues computer software. One approach for combating unreliability in software is through study of failure data collected from developmental and operational software projects. A system for comprehensive collection of interesting software events is vital to the accumulation of failure data that may be helpful in understanding software reliability problems. This thesis prescribes a well-defined data base of information about software events as an approach to identifying and understandinging aspects of unreliable software.

The focus of this work is primarily upon the question of what data are to be collected and the method for collecting these data. The nature of software errors is discussed with emphasis on characteristics and relationships to reliability. A summary description of selected software reliability models and an explanation of the kinds of data that are required to use these models is provided.

The Software Events Tracking System (SETS) is described in detail. This program makes possible the collection of critically needed data about software performance and changes to software. This data base will promote greater insight into the error phenomena of modern software systems. This insight will lead to better understanding of the error phenomena and to the development of new tools for increasing the reliability of software.