Soft-LLFI: A Comprehensive Framework for Software Fault Injection

The fact that many software systems are still plagued by critical software bugs conducted researches to deal with methodologies which introduce bugs into the software and recover from them. Since such methodologies need to be well-tested, it necessitates using fault injection techniques to prove their effectiveness. Software Fault Injection (SFI) is defined as the process of deliberately injecting faults into software system in a controlled manner to observe their behavior in real time in the presence of faults. However, to achieve the conclusive result of SFI experimentation, it is crucial that SFI applies the fault models truly manifesting the real fault scenarios of interest. This paper makes the following contributions to the current state of the art: to propose 1) a comprehensive reference of software fault models faithfully representing a wide spectrum of investigated software bugs in real applications, and 2) a robust configurable SFI architecture to automate injection of software fault model of appropriate type, and to visualize impact analysis.