Biomedical and Civil Engineering Experiences Using Grid Computing Technologies

J M Alonso, V Hernández, and G Moltó. Biomedical and Civil Engineering Experiences Using Grid Computing Technologies. In Parallel Computing (ParCo) 2005, pp. 647–654, 33, NIC-Directors (John von Neumann Institute for Computing), 2005.

Download

[198.2kB pdf]  [HTML] 

Abstract

In this paper we describe the usage of Grid Computing technologies in two applications belonging to different scientific fields. The first one involves the 3D dynamic analysis of large-scale buildings. The second one performs electrical simulations of cardiac tissues. In both applications, an iterative simulation procedure gives place to a computationally and memory-intensive process. Those two previously implemented applications have been ported to a Grid infrastructure by developing and using a generic software layer called GMarte, which simplifies the process of executing a scientific software on a Globus-based Grid infrastructure. To test the benefits of Grid Computing in the applications considered, executions have been performed both on a local and a large-scale Grid infrastructure. On the one hand, the local infrastructure is composed of three clusters of PCs from our local research group. On the other hand, we have also employed part of the computational resources available in the framework of the EGEE project, the largest distributed deployment currently available for e-science. In both cases, reductions in the global execution times of the case studies have been achieved by performing multiple concurrent executions on the distributed infrastructures. A description of GMarte, the middleware used, is provided along with a discussion of important problems that must be considered when employing a large Grid infrastructure.

BibTeX Entry

@inproceedings{Molto2005bce,
   abstract = {In this paper we describe the usage of Grid Computing technologies in two applications belonging to different scientific fields. The first one involves the 3D dynamic analysis of large-scale buildings. The second one performs electrical simulations of cardiac tissues. In both applications, an iterative simulation procedure gives place to a computationally and memory-intensive process. Those two previously implemented applications have been ported to a Grid infrastructure by developing and using a generic software layer called GMarte, which simplifies the process of executing a scientific software on a Globus-based Grid infrastructure. To test the benefits of Grid Computing in the applications considered, executions have been performed both on a local and a large-scale Grid infrastructure. On the one hand, the local infrastructure is composed of three clusters of PCs from our local research group. On the other hand, we have also employed part of the computational resources available in the framework of the EGEE project, the largest distributed deployment currently available for e-science. In both cases, reductions in the global execution times of the case studies have been achieved by performing multiple concurrent executions on the distributed infrastructures. A description of GMarte, the middleware used, is provided along with a discussion of important problems that must be considered when employing a large Grid infrastructure.},
   author = {J M Alonso and V Hernández and G Moltó},
   editor = {G R Joubert and W E Nagel and F J Peters and O Plata and P Tirado and E Zapata},
   booktitle = {Parallel Computing (ParCo) 2005},
   pages = {647-654},
   publisher = {NIC-Directors (John von Neumann Institute for Computing)},
   title = {Biomedical and Civil Engineering Experiences Using Grid Computing Technologies},
   volume = {33},
   url = {http://www2.fz-juelich.de/nic-series/volume33/647.pdf},
   year = {2005}
}