The theory and practice of runtime reflection

Mehr zum Buch

As interactions and dependencies within distributed reactive systemsincrease, the problem of detecting failures which depend on the exactsituation and environmental conditions they occur in grows. As aresult, not only the detection of failures is increasingly difficult,but also the differentiation between symptoms of a fault, and theactual fault itself. This work proposes an efficient approach for thedynamic analysis of such systems. It introduces a framework, referredto as runtime reflection framework, for the detection of failures aswell as identification of their causes. Failure detection is basedupon monitoring systems w.r.t. their properties, defined in a customhigh-level specification language. Properties are translatable intotemporal logics for which efficient monitors can be generated. Basedon the monitors'' verdicts, a dedicated failure diagnosis is performedto identify explanations for an observed deviation. Either thisconfirms that a monitor detected the root cause for a failure, orindicates that the fault is located elsewhere. This work developsboth the theoretical foundations for runtime reflection as well asefficient means for its implementation.