GTBW (Gigabit Testbed West)


Topic: Gigabit Testbed West
Granted by: BMBF
DFN-Verein
Grant Period: 01.08.1997 - 31.01.2000
Contact Person: Dr. Thomas Eickermann
Partners: Forschungszentrum Jülich GmbH, Zentralinstitut für Angewandte Mathematik
GMD - Forschungszentrum Informationstechnik GmbH, St. Augustin
Deutsches Klimarechenzentrum
Alfred-Wegener-Institut für Polar- und Meeresforschung
Pallas GmbH, Brühl
Kunsthochschule für Medien Köln
Zentrum für Paralleles Rechnen (ZPR) der Universität zu Köln
Deutsches Zentrum für Luft- und Raumfahrt (DLR)
echtzeit GmbH
Rheinische Friedrich-Wilhelms-Universiät Bonn, Institut für Angewandte Mathematik


Synopsis

Communication networks with bandwidths of Gigabits per second (Gbps) are required wherever high-performing sources and drains of data have to be interconnected. In scientific computing areas with a considerable demand for high communication bandwidths include: coupling of supercomputers (meta-computing), high-resolution visualization, and distributed access to huge amounts of data.

A private ATM link between Jülich and St. Augustin with a bandwidth of initially 622 Mbps (and 2.4 Gbps since August 1998) is used for research projects in all of the above mentioned areas. Besides that, the network itself is subject to further investigations. In 1999 the ATM link was extended to connect several research institutions in Cologne and Bonn to the testbed.

Goals

The project was set up to demonstrate the usefulness of high-speed wide-area communication networks for scientific computing. A couple of selected applications which are known to need communication bandwidth of Gbps are investigated to achieve this aim.

The major objective is the coupling of architecturally different supercomputers - vector computers and massively parallel computers - to set up a heterogeneous meta-computer. Selected applications from geochemistry, medicine, climate and weather-modeling, high-resolution visualization, and industrial codes (CISPAR) that could benefit from such a configuration will be ported to the meta-computer. The experiences made here should strengthen the know-how needed for the planned coupling of the German supercomputer-centers.

The project sets up the first Gigabit wide area network in Germany, so the investigation of the network components is another important goal.

Subprojects