|
Abstracts
|
Franck Cappello
CGP2P project—a
logiciel project supported by French Minister of Research - Slides
" Investigating
the impact of the Large Scale on distributed systems "
The
scale of a distributed system has a consequence on the approaches that
could be used for implementing its basic features such as its global
organization, the Scheduling, Communication and Storage functionalities.
At large scale more autonomy should be given to the nodes implementing
them. In this short talk we present four of our research projects investigating
the impact of the large scale in the design of distributed systems:
1) XtremWeb
is an effort toward a P2P system for high performance computing (pull
mode + sandbox),
2) MPICH-V is research effort developing several protocols
for a fault tolerant MPI implementation Dedicated to large clusters,
Desktop Grids and Grids (uncoordinated checkpoint),
3) SimLargeGrid is simulator of fully distributed scheduling
algorithm (nearest neighbour scheduling).
4) Grid Explorer is a very large cluster (1K CPU) dedicated
to the simulation/emulation of
Grid
and P2P systems (reproducible experimental conditions).
Most
of these projects have been supported or initiated during the ACI GRID
project called CGP2P.
|
Christine Collet
“A
mediation framework for a transparent access to largely distributed data
sources”
Mediation
systems have been proposed to provide access to data distributed all
over the net and stored in sources with different formats and data
models. Such systems generally provide a global view of the underlying
systems and hide several functions on data, queries and transactions
for mediating the underlying systems. When looking at heterogeneous and
distributed biological data sources, existing mediation systems provide
a minimal infrastructure but none of them is able to represent characteristics
of sources such as functional dependencies or semantic equivalence, nor
to optimise queries dynamically.
The
mediation framework we proposed will allow to build systems able to:
(i)
consider sources containing weakly structured data which are generated
by applications and stored as HTML or XML files and generate mediation
queries over these sources;
(ii) authorise partial results for queries in case of data sources unavailability;
and (iii) be efficient even if queries are complex and/or net traffic
slows down.
The talk will focus on two aspects of the framework: (i) mediation queries
generation
(ii)
iterative and dynamic query evaluation. It will also show how such functionnalities
are interesting for biological mediation systems. |
Frédéric Desprez/Jean-Yves
l'Excellent
Activities
around Client-Server Computing over the Grid -
Slides
Contact:
desprez@ens-lyon.fr
Laboratoire de l'Informatique du Parallélisme
UMR 5668 CNRS-INRIA-ENS Lyon
ENS Lyon, 46, allée d'Italie, 69364 Lyon cedex 07
Téléphone : +33 (0)4 72 72 85 69
Fax : +33 (0)4 72 72 80 80
“Activities around client-server computing over the Grid”
Several
environments, usually called Network Enabled Servers (NES), have developed
such a paradigm: NetSolve, Ninf, NEOS, OmniRPC, and more recently DIET
developed in the GRAAL project. A common feature of these environments
is that they are built on top of five components: clients, servers,
databases, monitors, and schedulers. Clients solve computational requests
on servers found by the NES. The NES schedules the requests on the
different servers using performance information obtained by monitors
and stored in a database.
Designing
such a NES implies to address issues linked to several well-known research
domains:
·
scheduling to allow clients to chain requests in a workflow mode,
·
middleware and application platforms as a base to implement the necessary
“glue” to broke clients requests, find the best server available and
then submit the problem and its data,
·
distributed algorithms to manage the requests and the dynamic behaviour
of the platform.
Finally, “classical”
parallelism is used at the server level and between servers
|
Bruno Levy
GeoGRID project—a
multidisciplinary project supported by French Minister of Research
- Slides
Contact :Bruno.Levy@loria.fr
LORIA : UMR 7503
Campus scientifique BP 239
F-54506 Vandoeuvre-lès-Nancy Cedex France
Tel: +33 (0)3 83 59 20 77
“Distributed Geomodeling”
Oil exploration
involves a wide spectrum of expertises, spreaded accross a large number
of people located in different areas over the world. In this context,
the main challenge is to share not only computational resources, but also
information. In other words, the geomodeling workstation needs to becomes
a portal to a network, enabling both computational power and databases
to be shared. We present new solutions for distributed and cooperative
visualization. |
Johan Montagnat
MEDIGRID project—a
multidisciplinary project supported by French Minister of Research
- Slides
Contact:
johan@creatis.insa-lyon.fr
Creatis, INSA, bât. B. Pascal
69621 Villeurbanne Cedex France
http://www.creatis.insa-lyon.fr/~johan/
“Computing and data
grids for medical applications”
Digital medical
data represent tremendous amount of information for which automatic processing
is increasingly needed. This talk will present some challenges that are
facing medical image analysis applications in terms of volume of data
to manipulate and of heterogeneity of data. The basic requirements to
build future medical applications will be analyzed. Grid infrastructures
can bring the necessary processing and storage capabilities, but the existing
implementations have to be adapted to medical data constraints, especially
in terms of security and semantic content of the data. We will present
on-going work supported by the French Ministry of Research. |
Thierry Priol
Contact:
Thierry.Priol@inria.fr
INRIA Unité de recherche de Rennes
Campus universitaire de Beaulieu
35042 Rennes Cedex France
http://www.inria.fr/personnel/Thierry.Priol.fr.html
“Research
activities in Grid middleware in the PARIS research group”
The PARIS
project-team at INRIA-Rennes aims at contributing to the programming of
parallel and distributed systems (Grid infrastructures) for large scale
numerical simulation applications. Its research activities related to
Grid Computing concern: the design of a grid-aware software component
platform and an associated runtime to support the execution of grid-aware
components, distributed memory management within a grid using a P2P approach
and services for Grid computing within commodity operating systems. |
Guillaume Alleon
Slides
Contact: guillaume.alleon@eads.net
“”
|
Philippe Bricard
Grid Computing
Executive EMEA - Slides
Contact:
bricard@fr.ibm.com
Grid Computing Executive EMEA
Tel : +33-467-346-244
“Grid
Computing : Creating Information Technology and Business Value”
Today more than ever,
government agencies rely on the ability to secure, integrate and access
information across agencies and departments. In response to the informationaccess
challenges, government agencies are redefining their information methodologies
and retooling their IT infrastructures to make themselves more effective
in the new political environment. Grid computing can help manage largescale
data sharing and collaboration. This new approach to distributed computing
helps maximize use of existing data resources and makes both structured
(database) and unstructured (file-based) data available across a department
or organization. In addition, Grid computing can help government agencies
to secure data access and optimize storage.
These are just illustrations of what Grid Computing is all about. In this
presentation, we will discuss why and how Grid Computing is a key enabler
to deliver better efficiency through some specific examples of real life
projects. We will also exemplify the business case of Grid Computing for
IBM Corporation. |
Wolfgang Boch
Grid Technologies
for Complex Problem Solving - Slides
Contact:wolfgang.boch@cec.eu.int
European Commission, DG Information Society-F2
Head of Unit Grid Technologies for Complex Problem Solving
http://www.cordis.lu/ist/grids
Tel.: +32.2.296.35.91 Fax.: +32.2.299.17.49
“EU
Grid Research and the European Research Area”
An overview of the
research priorities and the objectives on Grid Research actions established
by the European Community as part of the Information Society Programme
of the EU's 6th Framework Programme for Research and Development (2002-2006)
will be presented. The presentation will include a brief analysis on national
activities on Grid Research as well as an introduction to policy challenges
and needs for specific actions for improved co-ordination between national
and Community initiatives. |
Antoine Petit
Direction Ministry
of Research - Slides
Contact: antoine.petit@recherche.gouv.fr
Ministère délégué
à la Recherche et aux nouvelles Technologies
Direction de la Recherche
1 rue Descartes F-75231 Paris Cedex
Tel: +33 (0)1 55 55 97 44 Fax: +33 (0)1 55 55 96 47
“The talk will
provide an overview of the French initiatives concerning GRID computing
and discuss of possible joint actions with UK. ”
|
Tony Hey
The UK e-Science
Programme - Status and Future Plans - Slides
Contact: Tony.Hey@epsrc.ac.uk
UK e-Science Programme Director
The talk will review
the achievements of the £250M 5 year UK e-Science Programme in
its first two years. After a review of the application pilot projects
the role of the Core Programme will be discussed. This will include
details of the UK e-Science Grid and the support activities. The engagement
of UK industry with the DTI element of the programme will also be highlighted.
The talk will conclude with a look to the future plans including details
of the proposed UK Open Middleware Infrastructure Institute and the
Digital Curation Centre.
|
Ron Perrot
GridCast : using
the grid in Broadcast infrastructures - Slides
Contact:
r.perrott@qub.ac.uk
Director, Regional e-Science Centre,
Queen's University,
Belfast
Gridcast is a research
project being undertaken by the BBC and the Belfast Regional e-Science
Centre. The main aim is to develop techniques for broadcasters to share
distributed media files and other distributed technical resources -
building on technical solutions already developed for computing grids.
The BBC's Broadcast network faces many of the problems which computing
grids are designed to solve. It is a distributed network, with processing
of broadcast material taking place at many nodes, carrying material
(broadcast video) which has very high bandwidth requirements and mixes
'live' and stored events. The BBC also has high demands for reliable
and consistent performance from its systems and networks. Special purpose
broadcast processing equipment (e.g. video editing suites or image rendering
devices) are located at specific points in the network but, in general,
are not available for use outside these locations..
The GridCast project is investigating the application of technology
developed for sharing distributed data files and computing resources
in a computing environment to the broadcast world. If successful it
will allow the BBC to plan for a future broadcasting environment where
broadcast material, and dedicated broadcast resources, can be shared
much more easily in the BBC's distributed environment.
|
David Snelling
Fujitsu Laboratories
of Europe - Slides
Contact:
d.snelling@fle.fujitsu.com
Fujitsu Laboratories of Europe
Hayes Park Central - Hayes End Road
Hayes, Middlesex UB4 8FE
+44-208-606-4649 (Office) +44-208-606-4539 (Fax) +44-7768-807526 (Mobile)
Progress Toward
Open Grid Middleware
Fujitsu
have been developing middleware for Grid computing for nearly a decade.
This contribution has been in the form of an implementation of the open
standard published by the Unicore Forum. In the past two years a wider
community has recognized the potential of the Unicore specifications
and simultaneously, there has been a global convergence toward the Open
grid Services Architecture. This talk will provide an update on this
convergence.
|
Mike Dewar
MONET
- EU FP5 Project - Leader, XML Technologies Group - Slides
Contact:
miked@nag.co.uk
NAG Ltd.
Wilkinson House - Jordan Hill Rd - Oxford OX2 8DR - UK
http://monet.nag.co.uk
Describing
and discovering mathematical web and grid services
While
WSDL and its extensions provides an effective way of describing the
interface to a web or grid service, it does not support a mechanism
for describing what the purpose of the service actually is. Within the
MONET project we have been developing such a mechanism - the Mathematical
Service Description Language (MSDL) - and investigating how we can use
semantic web technologies such as RDF and OWL to reason about the suitability
of a particular service for solving a particular problem. The ultimate
goal is to be able to deploy a broker which, given a collection of service
descriptions written in MSDL and a user's problem, can construct one
or more plans for solving the user's problem using the available services.
|
John Barr
Grid Computing
Specialist Sun Microsystems - Slides
Contact:
john.barr@sun.com
Sun Microsystems Ltd
Guillemont Park
Minley Road, Blackwater - Camberley GU17 9QG
Phone: 01252 421157 Fax: 01252 420105 Mobile: 07808 328351
"Sun
and Grid Computing"
* Sun's involvement
in the UK eScience programme
* Sun's Grid collaboration in other projects
* Brief overview of Sun Grid products, services & solutions
* Key issues to be addressed in moving Grid from academia to industry
|
Malcolm Atkinson
Malcolm Atkinson,
Director, National eScience Centre - Slides
Contact:
mpa@dcs.gla.ac.uk
Malcolm P Atkinson Department of Computing ScienceUniversity of Glasgow
Glasgow G12 8QQ
Tel : +44 141 330 4359 Fax : +44 141 m330 4913
URL : http://www.dcs.gla.ac.uk/people/personal/mpa/
"Data, Data Everywhere And No Time To Think"
The challenge
of making effective use of globally distributed, diverse and growing
data collections is central to e-Science. Whilst the potential for discovery
from integration is obvious the technical and socio-economic hurdles
are growing. Essentially we are overwhelmed with opportunities and the
resources to exploit those opprtunities cannot possibly grow as fast
as demand. New strategies are needed, but they depend on adequate, robust
and flexible infrastructure. They also depend on standard frameworks
and systematic resource description. The projects at NeSC are first
steps. They are intended to pioneer the strategy and the technologies
that support it.
|
David Walker
Professor David
W. Walker
Director of the Welsh e-Science Centre - Slides
Contact:
david@cs.cf.ac.uk
School of Computer Science
PO Box 916
Cardiff University - Cardiff CF24 3XF
United Kingdom
"Grid-Related Research at the Welsh e-Science Centre"
This talk
presents an overview of e-Science and Grid research being conducted
at the Welsh e-Science Centre. Our research focuses on the following
areas:
1. The development of Grid middleware, in particular:
a. Workflow tools and description languages
b. Grid execution environments
c. Quality of service frameworks
d. Provenance and other metadata issues
2. The development of problem-solving environments (or portals) to provide
a high-level interface for Grid applications and to mediate access to,
and the administration of, Grid resources.
3. The development of collaborative Grid-aware visualisation environments
4. The development of Grid applications and the deployment of services
in Grid environments. Applications areas include bio-diversity, computational
electromagnetics, and engineering design.
|
Dave Pearson
OGSA-DAI - Open
Grid Service Architecture - Data Access and Integration - Slides
Contact:dave.pearson@oracle.com
Programme Manager OGSA-DAI
Oracle Corporation UK Ltd
Oracle Parkway
Reading, Berkshire RG6 1RA, UK
http://www.oracle.com
"
Data Access and Integration in a Grid environment"
OGSA-DAI
is a collaborative programme of work involving the Universities of Edinburgh,
Manchester and Newcastle, with industrial participation by IBM and Oracle
to develop Grid middleware for data access and integration. It is primarily
intended for the UK e-Science community to develop Grid based applications
which support collaborative working and information based problem solving.
OGSA-DAI also represents a significant contribution on behalf of the
UK e-Science Core Programme to extend the Grid model to include database
interoperability. The middleware functional definition will ultimately
form the basis of standards recommendations on Grid data services put
forward by the Database Access and Integration Services Working Group
to the Global Grid Forum (GGF). This presentation outlines the motivation
and requirements for the OGSA-DAI work. It describes the architectural
framework and the functionality of OGSA-DAI components and illustrates
the behaviour of Grid data services within the current framework. The
first version of OGSA-DAI software was released in January 2003. This
has been followed by to further major functionality releases with more
than 1000 downloads being taken. The software and associated documentation
is available under an open source license agreement at the following
site: http://www.ogsadai.org.uk/
|
John Brooke
co-Director eScience
North West ESNW
Realistic modelling of complex systems on Grids. - Slides
Contact:j.m.brooke@man.ac.uk
ESNW
University of ManchesterOxford Road,
Manchester, M13 9PL, UK
http://www.sve.man.ac.uk/General/Staff/brooke
"Realistic
modelling on a high-performance Grid"
In order to model complex
systems in a realistic manner we need to coordinate high-performance
computing and visualization to compare simulations with data from experiments.
The data can be streamed from and experimental apparatus directly or
previous experimental data can be accessed via databases. This involves
several Grid challenges which are being tackled in UK projects such
as Reality Grid (funded by EPSRC) and GODIVA (funded by NERC). The particular focus of Reality Grid
is the investigation of mesoscale modelling in oil extraction and modelling
of dynamic interaction of large biological molecules. The particular
focus of GODIVA is the investigation of oceanographic models and the
use of data assimilation from observations. Both projects have strong
industrial participation from Schlumberger, Edward Jenner Institute
for Computational Vaccinology (Reality Grid) and British Marine Informatics
(GODIVA).
The ability to steer running computations to explore
parameter space via the expertise of the research team.
2) Link high-performance computing and visualization and distribute
the output to multiple participants.
3) Define standards for steering resources of many different
types on a Grid.
4) Utilise commodity hardware for visualizing very large datasets.
Performance monitoring and migration of simulations running on Grids
This work is also being extended to European HPC Grids
such as EUROGRID and DEISA
|
Steven Newhouse
Technical Director
London eScience Centre - Slides
Contact:Steven
Newhouse
http://www.lesc.ic.ac.uk/
Department of Computing
South Kensington Campus
Imperial College London
SW7 2AZ
United Kingdom
"Building
a Marketplace for Grid Services"
The Grid
is evolving into a standards driven service oriented architecture
that is built upon commercially proven technologies such as web services.
While such an infrastructure provides the mechanisms to promote interoperability
there is still no financial incentive to provide these services. The
UK's Computational Markets project is developing such an enabling infrastructure
that will enable the flexible pricing of services and provide mechanisms
for the service provider to be paid by the service consumer. It is envisaged
that such an infrastructure will promote new business models and markets
for the global community.
|
Peter Dew
Chair of the White Rose Grid Consortium
- Slides
Contact:dew@comp.leeds.ac.uk
tel: +44 (0)113 343 5432
http://www.wrgrid.org.uk/
Institute of Informatics School of Computing University of Leeds
"The White Rose Grid: Practise and Experience"
This talk discusses the practise and experiences
in building White Rose Grid production that underpins a broad range
of e-Science projects (e.g. DAME, gViz, and Grid-enabled MRI) carried
out by the three Yorkshire Universities’ researchers jointly with their
industrial partners. The project successfully adopts a Virtual Organisation
model (VO) in order to advance larger scientific and engineering projects
by applying collective knowledge, expertise and skills base. This paper
raises practical issues pertaining to the Grid built with a joint effort
between Computer Scientists and Computer Services. The paper shares
our technical and application experiences. The e-Science pilot DAME
(Distributed Aircraft Maintenance Environment) will be used as the main
case study to illustrate the issues involved in building a multi-site
grid application for a significant application.
|
Alan R.B. Gould
BAE SYSTEMS Advanced Technology Centre
- Slides
Contact:
Alan.Gould@baesystems.com
Alan R.B. Gould
Group Leader, Integration Technologies, BAE SYSTEMS Advanced Technology
Centre,
Sowerby Building,
FPC267, PO Box 5, Filton, BRISTOL BS34 7QW. U. K.
Phone: +44(0)117 302 8258 Fax: +44(0)117 302 8007
""
The BAE SYSTEMS Advanced Technology
Centre is currently constructing a prototype Grid infrastructure - “BAEgrid”.
BAEgrid is designed as an in-house “laboratory” in which we can exercise
grid concepts, capabilities and tools to develop and evaluate a range
of Grid-enabled business scenarios. This is facilitated by our close relationship
with the world-leading UK e-Science programme, and is being implemented
in collaboration with several major hardware and software vendors. Grid
technology in its current form is ready to be deployed in areas of compute-intensive
activity and can demonstrably improve efficiency of deployed IT assets
within a single organisational domain. However, the big pay-off from Grid
is in the collaborative “Virtual Organisation” area. Here we are engaging
with several e-Science projects to test existing capabilities, and gain
insight into the requirements for future work. We will provide a brief
overview of the current status of the BAEgrid, and some of the collaborations
now running. Our experience demonstrates the importance of the involvement
of the end-user community in the development of Grid. We identify three
topics as crucial to the success of Grid in a business environment: security
(the fundamental enabler), semantics (the key to interoperability at the
applications level) and human factors (central to effective collaboration). |
Rob Smith
Technical Director, North East Regional
e-Science Centre - Slides
Contact:
rob.smith@ncl.ac.uk
tel: +44 (0)191 222 7827
fax: +44 (0)191 222 8232
m: +44 (0)797 468 5015
"Human Factors and virtual organisations"
Grid technology can be used to eliminate organisational
boundaries by providing mechanisms that enable virtual organisations
(VO's) to seamlessly share resources.
This offers the potential for new organisational structures and value
chains, with highly dynamic virtual organisations purchasing resources/services
and creating business relationships on-demand. This potential is limited
by trust. NEReSC is investigating the human and organisational factors
associated with trust in Grid-based virtual organisations in order to
establish a basis for realising the potential of Grid computing.
|
