The most prevalent software reliability practices rely on defect prevention during the development phase and failure detection and defect repair during the testing phase. However, despite the best plans of a software reliability program, a large number of failures still occur after product deployment. Software fault tolerance techniques are rarely implemented due to their complexity and resource requirements. A practical approach is to identify only key sections of code that will likely benefit the most from fault tolerance techniques. This can be accomplished by first analyzing historical defect data for the most common causes of software faults that result in system failures. Then, sections of code matching the criteria for these types of faults can be identified. Finally, applicable fault tolerance techniques can be evaluated for each section of targeted code. This approach will ensure that code complexity growth due to the addition of fault tolerance logic is limited, yet highly effective.

Key Words: Software Reliability, Software Fault Tolerance, Code Reviews, Software Defects, Conclusion