Mining precise performance-aware behavioral models from existing instrumentation

摘要: Software bugs often arise from differences between what developers envision their system does and that actually does. When faced with such conceptual inconsistencies, debugging can be very difficult. Inferring presenting accurate behavioral models of the implementation help reconcile view reality improve quality. We present Perfume, a model-inference algorithm improves on state art by using performance information to differentiate otherwise similar-appearing executions remove false positives inferred models. Perfume uses system's runtime execution logs infer concise, precise, predictive finite machine model describes both observed have not been but likely generate. guides inference process mining temporal performance-constrained properties logs, ensuring precision model's predictions. describe demonstrate how it over art.