Résumé
Data intensive applications are present in a wide range of domains such as computational science, multimedia signal processing and defense systems. Their particular characteristics in terms of high number of communication operations and data manipulations lead to the need of well-adapted design formalisms. This chapter advocates the combination of a repetitive structure modeling formalism and abstract clocks to specify data intensive applications. The resulting specifications consist of an expression of the potential parallelism inherent to described data-intensive algorithms, refined with environment constraints in the form of application component's activation rates. The modeling concepts used here are already integrated to the UML Marte standard profile, which allows for a graphical modeling and analysis of realtime and embedded systems.