A Taxonomy-based Approach for Fault Localization in Service-Oriented Applications

Abstract

Service Oriented Computing is a relevant paradigm that allows for building software solutions as it supports the creation of dynamic and agile applications. As a result of these benefits, applications based on the paradigm are commonly used in the industry. However, the more complex software solutions become, the higher the chance of having bugs is. Identifying them in the entire solution may also be one of the most tedious, expensive and time-consuming activity within the developing process. Therefore, fault localization aims at finding the location of faults, determining the root cause of the failure and identifying the causes of abnormal behavior of a faulty program. In particular, fault localization for service oriented applications is currently a research field that has received little attention. In this context, we present a taxonomy-based approach for fault localization in service oriented applications, using information of faults in the underlying architecture of these applications, namely SOA. The results on the location of faults are not only provided to the developer but also enhanced with architectural information, that is, showing the faults and their connection with exceptions, code and architectural scenarios. The experimental results carried out indicate that developers reduce their effort of localizing faults in terms of time and lines of code.