Evaluating Frameworks Assemblies In Microservices-based Systems Using Imperfect Information

Abstract

Microservices is an architectural style that promotes the facility to build and maintain systems by breaking down its business capabilities into smaller and distributed services. Often, practitioners commonly use frameworks to provide generic functionalities to address recurring quality attribute concerns on microservices-based systems. Nevertheless, in practical settings, frameworks information is incomplete, imprecise, and changing as well as requirements. More realistically deployable approaches combine the exploration of candidate architectures with their evaluation regarding requirements satisfaction and the fuzziness and incompleteness available frameworks information. This article outlines a novel technique, called $\mu$Azimut, whose purpose is to generate, evaluate, and compare frameworks assemblies using potentially incomplete, imprecise, and changing descriptions of non-functional requirements and frameworks. The frameworks assemblies evaluation is based on a support score which allows modeling imperfect architectural knowledge. The technique is evaluated in an industrial case study. The results point out that $\mu$Azimut generates solutions that are close to those solutions that an architect selects for designing microservices architectures.