ABSTRACT:

This article presents a model and methodology for deriving the expected financial consequence of software failure in the subsystems, or modules, of a complex computer system. The model incorporates assessments of the magnitude of errors, as well as inherent and control risks. First, hazards, threats associated with failure of a computer-based system, are identified by studying the environment in which the software operates. Procedures are included to assess the expected consequence of hazards, the potential economic loss and/or misstatements of the accounting data base that may occur when the software has errors and controls are ineffective. Module financial exposure, the consequence of errors in individual modules, is determined by relating the use of a module and the functions that it performs to the hazards and their expected consequences. The expected number of module failures is derived from a software reliability model. These estimates are combined to assess the expected module financial consequence of failure resulting from faults in individual modules of a software system.

A case study illustrates how the model and methodology may be used to assess the risk of material software errors and to focus testing on software modules identified as high risk. When applied to a commercial loan system installed at one of the 25 largest savings and loan associations in the United States, the model identified six of the 135 modules (four percent) as high risk and 11 modules (eight percent) as intermediate risk. These results indicate that only a few modules pose significant risk of contributing to adverse financial consequences. Awareness of these modules enables systems professionals to reduce the testing domain by concentrating on these relatively few high risk modules.