Open Access Open Access  Restricted Access Subscription Access

Using XML with Large Parallel Datasets: Is There Any Hope?

R. N. ELIAS, V. BRAGANHOLO, J. CLARKE, I. H. F. SANTOS, M. MATTOSO, A. L. G. A. COUTINHO

Abstract


The e-Science area has been calling attention of computer scientists to several challenges in supporting large scale scientific experiments. One of them is data representation. Providing scientists with efficient data access starts by choosing the best alternative to represent such data. An alternative concerned with data interoperability is necessary in such scenario. This paper evaluates XDMF as an XML-based alternative to represent data in large scale parallel CFD applications. In this format, data considered as “heavy” is stored in bi- nary files, while all information required to data access is stored in XML files, inheriting XML’s flexibility, interoperability between different applications, self-description and easyof- access. These advantages are demonstrated in a real case study, the interaction of a parallel coupled viscous flow and temperature solver with a parallel visualization system.

Full Text:

PDF