Towards Transparent Distributed Execution in the Tornado Framework Filip H.A. Claeys 1 , Maria Chtepen 2 , Lorenzo Benedetti 1 , Webbey De Keyser 1 , Peter Fritzson 3 , Peter A. Vanrolleghem 1,4 1 Department of Applied Mathematics, Biometrics and Process Control (BIOMATH), Ghent University, Coupure Links 653, B-9000 Ghent, Belgium fc@biomath.ugent.be 2 Department of Information Technology (INTEC), Ghent University, Sint-Pietersnieuwstraat 41, B-9000 Ghent, Belgium mchtepen@intec.ugent.be 3 Programming Environments Laboratory (PELAB), Link¨ oping University, SE-581 83 Link¨ oping, Sweden petfr@ida.liu.se 4 modelEAU, D´ epartement de g´ enie civil, Universit´ e Laval, Pavillon Pouliot, Qu´ ebec, G1K 7P4, QC, Canada peter@modelEAU.org Abstract. Tornado is a new advanced kernel for modelling and virtual experi- mentation (i.e., any evaluation of a model) in the water quality domain. Although primarily intended for use within this particular domain, the kernel is generic in nature and has a plethora of generally applicable features. Tornado often deals with elaborate models and many of its virtual experiment types are computation- ally intensive. In order to alleviate the computational burden, there is a strong need for distributed execution. However, since Tornado has a heterogeneous user community consisting of both expert and non-expert users, the distributed execu- tion process should preferably be as transparent as possible. This article focuses on the initial steps that were taken along the road to transparent distributed ex- ecution. Main achievement so far is the ability to perform semi-automated dis- tributed execution of workload on the Typhoon cluster and LCG-2 grid infras- tructures. Our approach is based on the generation of generic job descriptions and has been shown to offer sufficient transparancy for non-expert users in the scope of a Monte Carlo simulation project that was run on a 16-node Typhoon / 40-node LCG-2 setup. 1 Introduction In water quality research, the biological and/or chemical quality of water in rivers, sew- ers and wastewater treatment plants (WWTP) is studied. Research in this domain is facilitated by a number of models that have received a formal or de facto standardiza- tion status. Most notable are River Water Quality Model No.1 (RWQM1) [1] and the Activated Sludge Model (ASM) series [2]. Water quality models typically consist of large sets of non-linear Ordinary Differen- tial Equations (ODE) and/or Differential-Algebraic Equations (DAE). These equations