Reliability sensitivity analysis for Yahoo! Pipes mashups

Development of Web technologies has pushed forward a wide array of tools which can be used by the end-users to customize their own environment. Mashup development frameworks stand out as means to create Web applications by combining the existing content or functionality provided by various, disparate, providers on the Web. Like for all composite systems, reliability is one of the primary design concerns in mashup development. This is due to the fact that the content used in the mashup is often not under the supervision of the mashup developer and is rather deployed on the Web by a third-party provider. In this paper a reliability sensitivity analysis is conducted for the collected dataset of Yahoo! Pipes mashups. Reliability of each mashup in the dataset is estimated using a reliability model based on belief networks. A sensitivity analysis is conducted on the reliability models to ascertain which building components have the greatest impact on the overall reliability of the data set.