Performance of Object-Oriented Software Code: Profiling and Instrumentation

Pro ling is necessary for a better analysis of the software code performance. It consists of evaluating, at execution time, one or several metrics on the individual code components. In this paper, we present a scalable software pro ling model (SPM) that is a weak coupling of an exhaustive software code components structural model and a software metrics model. Moreover, we propose an automatic instrumentation method which is based on SPM and implemented on top of the O2 object-oriented database management system. Finally, the code components and their associated pro ling data must be stored in order to ensure the maintenance of the software. The structural and pro ling data are stored as persistent objects in a component repository according to SPM.