, Western University of Timisoara
Grid Computing for Electromagnetics is one of the few available books that presents real applications running on grids. It's a practical book for researchers who want to set up a computational grid and understand the perspectives that grid computing platforms offer. It would also be suitable as a classroom text for undergraduate physics or electrical engineering students with some knowledge of computer networks, C programming, and numerical analysis.
The book covers three topics: basic grid computing, grid middleware installation, and grid applications in electromagnetics.
The first section slowly introduces readers to grid computing, beginning with parallel, distributed, and Web computing issues, such as architectures, topologies, programming paradigms, and performance measurements. In this context, Luciano Tarricone and Alessandra Esposito present grid computing as the convergence of parallel, distributed, and Web computing. They discuss computational grids, data grids, and collaborative engineering under the poorly named section title "Computational Grids."
The authors don't introduce grid middleware until they inspect the enabling technologies. While most grid computing books highlight important security protocols, this book pays more attention to other enabling technologies, such as current implementations of object orientation. An appendix covers security issues.
This section focuses on the Globus Toolkit and only briefly mentions other grid-dedicated tools. The practical guide to building a grid follows the classical path: hardware and software requirements, downloads, installation, configuration, certificates, and available services and testing.
Tarricone and Esposito describe all the steps concisely, with concrete examples. One drawback is that the GT installation is particularly for version 2.2.4, while the current version is 4.0.
This section describes three grid applications: one for high-performance computing, one for collaborative engineering, and one for data management.
The first application is a C and MPI (message passing interface) implementation of the finite-difference time domain method, which is ported to a computational grid. The authors equally treat the basic algorithm, parallel-implementation issues, and benchmark results.
Concerning the collaborative-engineering application, Tarricone and Esposito construct a scenario in which they use a few proprietary software tools in a computer-aided-engineering simulation of aperture-antenna arrays. The book discusses numerical methods encapsulated in the software tools, revealing their need for high-performance computing platforms. The authors use Nimrod-G as a resource management tool for a Globus-based grid.
The third application concerns an integrated system for optimum-wireless-network planning. A sequence of simplifications leads to a real-time data-management model implemented on top of a Globus-based grid. The CD-ROM includes fragments of the application codes and the full Globus Toolkit 2.2.4.
Each chapter has a large bibliography. The glossary, appendices, and lists cover an overview of Linux, a short introduction to the mathematics behind electromagnetic theory, a short list of Web sites, and a welcome glossary of common grid computing terms.
Because of the book's pragmatic value, I'd recommend Grid Computing for Electromagnetics to graduate students and professionals who'd like to start their own grid-based application or become a grid expert. The authors integrate diverse topics and clear up some technical issues behind grids. It's easy to read and focuses on applications rather than theory, distinguishing it from other resources. Its helpful illustrations and code examples are strengths, but a clear disadvantage is its higher price compared to other books in the field.
Cite this article: Dana Petcu, "Parallel Numerical Applications on Grid Environments," IEEE Distributed Systems Online, vol. 6, no. 4, 2005.