Assessment of Software Reliability

This paper discusses methods for and problems in achieving reliability of large-scale software systems. Comparative studies were made of a U. S. Air Force software project, a NASA software project, and a commercial software project. Software development and test management procedures which lead to software reliability are analyzed. The underlying premise is that software reliability must be designed into the system from the outset using a systems approach. The systems approach to achieving software reliability requires (1) understanding of the total software development and test life cycle, (2) identification of conventional and extended conventional test techniques for precision validation testing of applications programs, and (3) allocation of resources in a cost- and performance-effective manner, in advance, over the entire development period.