Towards automatic compositional performance analysis of component-based systems
Vincenzo Grassi, Raffaela Mirandola · ACM SIGSOFT Software Engineering Notes · 2004
To make predictive analysis an effective tool for component-based software development (CBSD), it should be, as much as possible:compositional, to allow the re-use of known information about the properties of existing components, andautomatic, to keep the pace with the timeliness and cost-effectiveness promises of CBSD. Towards this end, focusing on the predictive analysis of performance properties, we define a simple language, based on an abstract component model, to describe a component assembly, outlining which information should be included in it to support compositional performance analysis. Moreover, we outline a mapping of the constructs of the proposed language to elements of the RT-UML Profile, to give them a precisely defined "performance semantics", and to get a starting point for the exploitation of proposed UML-based methodologies and algorithms for performance analysis.