Software Self-adaptation and Industry: Blame MAPE-K

Abstract

If software self-adaptation has to be widely adopted by industry, we need to think big, embrace complexity, provide easily deployed and cost-effective solutions, and justify trust. On fairness, MAPE-K should not solely take the blame. MAPE-K is one of the many interpretations of feedback loops apply to systems for which mathematical models - mostly based on control theory, are difficult to be synthesised. MAPE-K has provided a basic and widely accepted framework for justifying the deployment of feedback loops in software systems. Undoubtedly, it has helped to promote and advance the whole area, but now more concrete and resilient solutions are necessary. This position paper argues that, first, industry has been adopting software self-adaptation, perhaps in a way that may not be recognised by the academic community, second, generic solutions are unfeasible since every software system brings its own challenges, and thirdly, the generic stages associated with a feedback loop, like MAPE-K, are insufficient to solve specific challenges.