Middleware Adaptation through Process Mining

Abstract

The development of adaptive middleware systems is a complex task due to the difficulty of dealing with adaptation issues, such as 'how' to implement the adaptation mechanism, 'where' to insert the adaptive code into the middleware, and 'when' the adaptive code is composed with the middleware logic. Existing solutions to build adaptive middleware usually concentrate on the use of software technologies like aspect oriented programming and computational reflection to face with the how issue. In this paper, we propose a solution to build middleware that is adapted at runtime 'when', whose adaptation decisions and actions are moved from the middleware to an external component 'where' and whose adaptation makes use of process mining techniques and software architecture 'how'. The adaptation process is triggered based on the verification of the middleware event log. In order to evaluate the proposed approach, we carried an experimental evaluation to check the quality of the mined middleware model and the verification overhead.