Software Diversity—Some Considerations About its Benefits and its Limitations

Abstract Recent experimental results have suggested that the use of software diversity might not be the proper means to ensure safe operation of software with high reliability demands. This paper shows for realistic applications that the number of software failures of a two-fold diverse system that are in common to both versions is expected to be significantly smaller than the number of failures in only one such version. So back to back testing may still be an economic means to show low failure probabilities. A comparison is made whether it is more economic to use diversity or whether it is beneficial to use singular systems only.