Self-healing in Service Mashups Through Feature Adaptation
Mahdi Bashari, Ebrahim Bagheri, Weichang Du · 2017
The composition of the functionality of multiple services into a single unique service mashup has received wide interest in the recent years. Given the distributed nature of these mashups where the constituent services can be located on different servers, it is possible that a change in the functionality or availability of a constituent service result in the failure of the service mashup. In this paper, we propose a novel method based on the Software Product Line Engineering (SPLE) paradigm which is able to find an alternate valid service mashup which has maximum possible number of original service mashup features in order to mitigate a service failure when complete recovery is not possible. This method also has an advantage that it can recover or mitigate the failure automatically without requiring the user to specify any adaptation rule or strategy. We show the practicality of our proposed approach through extensive experiments.