View the presentation slides  (4 Mb)

Abstract

This presentation describes the technological platform used by the QosCosGrid consortium (partners in Europe and Australia) to enable compute-intensive complex system simulations to be implemented and the results to be shared across the research community.

The term supercomputer typically refers to a dedicated special-purpose multiprocessor computing system that provides close to best achievable performance for demanding parallel workloads. Grid computing systems could be viewed as large-scale computing systems with considerable levels of hardware resources but with a lack of the features that make supercomputers so powerful, e.g. parallel execution environments, scheduling or multi-user support.

The main challenge of the European research and development QosCosGrid project, in which University of Queensland (Institute for Molecular Bioscience) from Australia is actively involved, is to overcome the current limitations of grids and implement an international testbed together with a collaboration platform which will help both complex system developers and users work on large-scale computer-based simulations. Typically, such simulations require computing environments that meet very high specifications in terms of processing units, primary and secondary storage, and communication. Acquiring, operating and maintaining supercomputers involves considerable costs that many organizations cannot afford. Therefore, we have created a new multi-institution distributed system to provide application-driven features and performance characteristics that resemble those of a supercomputer.

QosCosGrid software components and development rules are based on open standards (e.g. taken from Open Grid Forum) as well as best practice and results from previous international research projects, e.g. GridLab, GridSphere/Vine and HPC-Europe. In order to protect sensitive data and computing resources we have adopted advanced authentication and access control mechanisms to define, control and enforce cross-administration policies on resources. As a proof of concept we have selected nine complex system use cases representing various scientific domains, including molecular interactions, cells and organisms, ecosystems, human economies, and modern energy and telecommunication infrastructure. Various scientists using the QosCosGrid system have not only managed to execute and control their cross-site compute-intensive simulations but also have established research collaborations and exchanged results from their experiments over the QosCosGrid web platform. The whole system and its components are based on open source and will be publicly available in Q4 2008.

About the speaker

Pamela Burrage is currently working on the QosCosGrid project through the ARC Centre of Excellence in Bioinformatics, Institute for Molecular Bioscience at the University of Queensland. This project (quasi-opportunistic supercomputing for complex systems in grid environments) is a major 6th framework EU project, with partners in Europe and Australia. Pamela has a PhD in Mathematics (from UQ), and has interests in mathematical modelling and the numerical solution of stochastic differential equations with applications in computational biology, as well as being interested in visualisation and parallel computing. She has 17 published journal articles, two published (refereed) conference papers, one article submitted and another article recently accepted for publication, in these fields of interest. Pamela was recently an invited speaker at the 2nd Bertinoro Systems Biology Workshop (April 2008, in Italy). Pamela’s role within QosCosGrid is to implement mathematical simulations on the plasma membrane, across a computational grid. Such simulations will aid in the understanding of the complicated dynamical processes that take place on the plasma cell membrane (both spatially and temporally).