Performance modeling and validation of a software system in a RT-UML-based simulative environment

The performance validation of software systems is becoming a crucial activity of the software development process. This is mostly due to the resource sharing and the remote deployment of software objects that may introduce critical delays in performance indices like the system response time. Hard and soft real-time systems are particularly affected from performance issues, therefore the ability to model and validate this attribute may become an extra value in software development environments. In this paper we introduce a framework to model performance aspects using the real-time object modeling (ROOM) notation. We devise a standard approach to represent hardware resources (such as CPUs and disks), to formulate resource requests of software objects, and to model the delays and the resource contentions that may arise from such requests. The integration of the software model and the resources is made transparent to the software developer by exploiting the integration features of the Rose Real Time (RRT) tool. RRT is based on the real time UML notation that is an implementation of ROOM. We also show an example of application of our framework on a video system case study