The variation of software survival time for different operational input profiles (or why you can wait a long time for a big bug to fail)

摘要

Experimental and theoretical evidence for the existence of contiguous failure regions in the program input space (blob defects) is provided. For real-time systems where successive input values tend to be similar, blob defects can have a major impact on the software survival time because the failure probability is not constant. For example, with a random walk input sequence, the probability of failure decreases as the time from the last failure increases. It is shown that the key factors affecting the survival time are the input trajectory, the rate of change of the input values, and the surface area of the defect (rather than its volume).