Visualization gViz - Middleware for e-Science

advertisement
gViz - Visualization Middleware for e-Science
Jason Wood, Mark Walkley, Chris Goodyer, Ying Li and Ken Brodlie - University of Leeds
The gViz project is a major part of the UK e-Science research programme, aiming
to provide today’s e-Scientist with visualization software that works within modern
Grid environments.
Grid-enabling Current Visualization
Systems
A major part of gViz is the Grid-enabling of existing
visualization systems, so that scientists can migrate their
work seamlessly to Grid computing environments – we
have extended a widely used visualization system, IRIS
Explorer from NAG Ltd. Our extension allows the
familiar dataflow network to span a set of Grid resources,
so that user interface modules execute on the scientist’s
desktop, but computationally intensive modules are
launched securely on remote servers using Globus
middleware. Moreover a number of scientists at different
locations can join in a collaborative visualization
session, using the IRIS Explorer COVISA modules.
Grid-enabled
IRIS Explorer
Modules in the
dataflow pipeline
execute on
different Grid
resources
Collaborative IRIS
Explorer
Geographically
separated research
teams collaborate
across the network
Grid-enabled Computational Steering
Computational Steering
Here the gViz library connects an environmental
simulation to a front-end visualization system.
Above, IRIS Explorer is used to provide the visual
interface – while below, SCIRun, from the
University of Utah, and Qt/VTK have been used.
A special focus of the gViz project is Computational
Steering. Visualization runs in tandem with simulation, and
the scientist can amend the controlling parameters of the
simulation as it executes. The gViz Computational Steering
Library allows scientists to link their simulation code –
running remotely - with a visualization system of choice –
running on the desktop. The Library can operate in a Web
Services context, with the opportunity to register simulation
details with a Web Service. These details may later be
retrieved by a visualization system that is used to visualize
the computed data and subsequently steer the simulation.
This library has been used in the e-Science project “Parallel
Grid Computing for Lubrication” presented on a separate
poster.
Our work is carried out within the White Rose Grid e-Science
Centre of Excellence. The White Rose Grid is a Service Grid,
providing high performance computing for the Universities of
Leeds, Sheffield and York.
Partners in the project are: Universities of Leeds, Oxford and Oxford Brookes; CLRC Rutherford Appleton
Laboratory; NAG Ltd; IBM UK; and Streamline Computing.
Further information at: http://www.visualization.leeds.ac.uk/gViz
Download