A Causal Model to predict the Effect of Business Process Evolution on Quality of Service

Description

Managing Quality of Service (QoS) of Service-based systems is a key challenge to produce systems that fulfill their requirements. Verifying the respect of a QoS contract in a system becomes more and more difficult as systems are more and more complex. Moreover, systems have to evolve in order to fulfil constantly changing requirements. As QoS properties are influenced by hidden factors such as connection rate or the system execution itself, determining the cause of a performance degradation is not mainstream. We propose in this paper to identify the causal relations to make explicit the hidden factors of influence. We more specifically focus on the consequences of system evolution with respect to QoS properties: using causal relations, we aim at predicting the possible overhead caused by an evolution. This paper shows through an example of Business Process how our evolution analysis helps to understand the effect of evolution on QoS property such as the Response Time. We show its efficiency by comparing the prediction with measured values.

Abstract

International audience

Additional details