Large-Scale Computing Techniques for Complex System Simulations

Chapter 9

QosCosGrid e-Science Infrastructure for Large-Scale Complex System Simulations

Krzysztof Kurowski, Bartosz Bosak, Piotr Grabowski, Mariusz Mamonski, and Tomasz Piontek

Poznan Supercomputing and Networking Center, Poznan, Poland

George Kampis

Collegium Budapest (Institute for Advanced Study), Budapest, Hungary

László Gulyás

Aitia International Inc. and Collegium Budapest (Institute for Advanced Study), Budapest, Hungary

Camille Coti

LIPN, CNRS-UMR7030, Université Paris 13, Villetaneuse, France

Thomas Herault and Franck Cappello

National Institute for Research in Computer Science and Control (INRIA), Rennes, France

9.1 INTRODUCTION

Grids and clouds could be viewed as large-scale computing systems with considerable levels of hardware resources but lacking many of the the features that make supercomputers so powerful. In particular, grids and clouds usually do not provide sophisticated support for parallel and multiphysics applications with significant interprocess communication requirements. Connected via local and wide area networks, such big computing infrastructures typically rely on an opportunistic marshalling of resources into coordinated action to meet the needs of large-scale computing applications. Both grids and clouds are often presented as a panacea for all kinds of computing applications, including those that require supercomputing-like environments. However, this vision of grids or clouds as virtual supercomputers is unattainable without overcoming their performance, ...