A Fault Injection Based Approach to Assessment of Quality of Test Sets for BPEL Processes

Abstract

Mutation testing is an effective technique for assessing a quality of test sets for software systems, but it suffers from high computational costs of generating and executing a large number of mutants. In the domain of BPEL processes each mutant needs to be deployed before it can be executed, thus the cost of processing mutants increases further. In contrast to mutation testing, fault injection is able to inject faults directly into the original process what reduces the redeployment requirement. The paper presents an experiment of the application of software fault injection to assess quality of test sets for BPEL processes. Faults are introduced by a Software Fault Injector for BPEL Processes (SFIBP). SFIBP simulates effects of the faults by modifying invocations of web-services and their internal variables. The experiment proved high superiority of the application of the SFIBP over the mutation testing, especially in the case of time requirements.