Failure Patterns and Reliability Growth Potential for Software Systems

Abstract

In the development of a complex software system the program manager is generally interested in the performance and reliability of the total system, which includes both the hardware and the software. This paper addresses a number of reliability characteristics and concepts for hardware and software which have proven useful in the development of reliable complex systems. These include reliability growth potential, deterministic-state and stochastic state computer systems, and sampling of the input space. The growth potential is the maximum reliability that can be attained with the system design and reliability management strategy. This paper discusses the utilization of the growth potential concept to evaluate the hardware reliability growth during development testing. The state of the computer is the total computer memory together with the logic step of the program, and may depend on the physical environment in which the system operates. Deterministic and stochastic state computer systems are defined and it is noted that the failure patterns for these systems are typically different. In addition, a method for assessing a software reliability parameter based on random sampling of the input space will be discussed and illustrated by example.