An Architectural Framework for Analyzing Tradeoffs between Software Security and Performance

The increasing complexity of software systems entails large effort to jointly analyze their non-functional attributes in order to identify potential tradeoffs among them (e.g. increased availability can lead to performance degradation). In this paper we propose a framework for the architectural analysis of software performance degradation induced by security solutions. We introduce a library of UML models representing security mechanisms that can be composed with performance annotated UML application models for architecting security and performance critical systems. Composability of models allows to introduce different security solutions on the same software architecture, thus supporting software architects to find appropriate security solutions while meeting performance requirements. We report experimental results that validate our approach by comparing a model-based evaluation of a software architecture for management of cultural assets with values observed on the real implementation of the system.