Formalizing and Deploying Grid-Based Applications Using a Formal Architectural Model-Driven Approach David Manset[1, 2, 3], Hervé Verjus[1], Richard McClatchey[2] [1] University of Savoie – Polytech’ Savoie – LISTIC/LS {david.manset, herve.verjus}@univ-savoie.fr [2] CCCS, University West of England, Bristol, UK {richard.mcclatchey@uwe.ac.uk} [3] Maat Gknowledge, Toledo, Spain {dmanset@maat-g.com } Keywords. Grid, MDE, Software Architecture, Architectural Properties, SOA, ADL, Model Transformation. Abstract. Nowadays, technological developments have led to the possibility of using distributed computing facilities as a single, unified computing resource, a so-called Grid, following the analogy of a power Grid providing consistent, per- vasive, dependable and transparent access to electricity, irrespective of its source or its delivery. Grids implement numbers of protocols, applications, standards and attempt to virtualize and harmonize accesses to those. Most implementations therefore result in superposing software layers, often resulting in very low qual- ity of services and quality of applications, making design and development in- creasingly complex, and leading most classical engineering practices to collapse. This paper advocates a formal engineering approach to Grid system develop- ments in an effort to contribute to the rigorous design of Grids software architec- tures. This approach addresses quality of service and cross-platform develop- ments by applying the model-driven paradigm to a formal architecture-centric engineering method. 1. Introduction Today, the Internet enables users worldwide to access multiple online resources. Its popularity, combined with the availability of powerful computing resources (including networks, storage devices, computers…) as low-cost commodities is changing the way we use computers. New technologies have led to the possibility of using distributed computing facilities as a single, unified computing resource, a so-called Grid, using the analogy of a power Grid that provides consistent, pervasive, dependable and trans- parent access to electricity irrespective of its source and delivery. Grids enable the ag- gregation, virtualization and sharing of massive heterogeneous and geographically dispersed resources, using files, applications and storage devices, to solve computation and data intensive problems, across institutions and countries via temporary collabora- tions called virtual organizations (VO) as described in (Foster et al., 2001). Simply