GridCoord DoW

Caveats

This report delivers only a compendium of the activities that are pursued in Universities and Research Labs in Europe. It is not meant to be complete, and it covers only the countries represented in GridCoord. Collected information is provided as reported by the project partners: it is not fully tested for accuracy. One GridCoord member, Sweden, did not provide the requested information and thus it is not covered.

Conclusions and recommendations

This report shows that the European countries have a prestigious research tradition in the areas that contributed “to build” the Grid technology - high-performance computing, parallel and distributed processing, networking infrastructures, software engineering in objects/components platforms, data base and knowledge discovery, cooperative computing -, as well as in the scientific application sectors: computational physics, high-energy physics, computational chemistry, life sciences, genomics, medicine, astrophysics, geophysics, earth observation systems, weather prediction, astrobiology, while some significant industrial applications are emerging: telecommunications (VHTD++, France, and VIOLA, Germany), drug industry (ADMEToxGrid, Hungary), Data mining (Data Mining Grid, Hungary), e-banking (SAIB, Italy) and supply chain management (SFIDA, Italy).

First of all, it must be pointed out that, though there have been important advances in basic technologies (such as infrastructures, programming interfaces and tools, scheduling and resource management techniques, monitoring, static analysis, and the such), much remains to be done in basic and applied research in Grid Computing for this technology to be successfully exploited at the industrial level: open middleware, operating systems, programming environments, composition of complex applications, security, data bases, vertical integration, as well as standardization.

The report shows that significant differences exist in the approaches that are currently applied in the various countries, each with its intrinsic strengths. This multinational character of Grid research in Europe is a valuable source of innovation. Nevertheless, a co-ordinated approach towards Grid research in Europe is essential to achieve critical mass, the potential for a more visible impact at an international level, and strategic industrial impacts.

We propose, therefore, to promote:

• 
  More coordination between national and European research on the one hand and between research and industry on the other hand;
  
• 
  Multilateral exchange of experiences, in order to cross-fertilise knowledge and expertise, by the identification of research and collaboration opportunities;
  
• 
  The deployment of a pan-European Experimental Grid platform, to give the opportunity to researchers of several countries to collaborate, to conduce large scale experiments, and be able to reproduce the ones conducted by others, to share ideas, results and developed software.
  

France

Summary of activities

Currently, the French activities on Grids are mainly focused on the Grid'5000 project. This project is described extensively in the next section. In a few words, it aims at building an experimental Grid platform on several sites, totalling close to 5000 processors (hence the name). It is funded by the French ministry of Education and Research, INRIA, CNRS, the Universities at the various sites and some regional councils, totalling 7.8 M€ (not including the salaries of the people, already paid by the State).

Grid'5000 is intended to provide a testbed for research in Grid Computing, eventually becoming the main tool for experiments for researchers in this particular field, as well as for those in connected disciplines. It is only for experiments, and it is not planned to be made available for production use. It is expected to be totally configurable, as users may install whichever piece of software they wish, from OS to high-end applications. Plans are being made to collaborate with existing large-scale grids, for example DAS-2 in the Netherlands or NAREGI in Japan.

The Grid'5000 project has been made possible thanks to a number of other key projects funded by the Ministry of Research and the Ministry of Industry: the VTHD/VTHD++ high-performance networking project, and the e-Toile grid middleware project. They are extensively described below.

Also, the long-term building of the French grid community has been instrumental in the success of Grid'5000. It has been mainly funded by CNRS through a series of long-term (4 years) collective research co-operation programmes (GDR) on architecture, networking, system and parallelism. These programmes have been funded for more than 15 years with an average level of 40 k€ a year in the latest years, and they federate essentially all the academic French groups active in this field. For a limited, but stable investment, they allowed to greatly increase the collaborations between the groups, to trigger researcher mobility at the junior and even senior levels, to disseminate all relevant information through mailing list, to serve as a “nursery” whence most if not all the current academic co-operation projects have grown up.

This GDR programmes have been able to set up stable series of French conference on various fields of parallelism, the eldest one being RenPar (for Parallel Encounters), which gathers around 150 (mostly junior) researchers each year together with other co-localized events. Also, both CNRS and INRIA have been every active in promoting national thematic schools on parallelism, high-performance computing and grids. The current state of the grid projects in France cannot be understood without a reference to this background community-building work. Also, again thanks to the decisive support of INRIA, CNRS and the Ministry of Research, the research groups have been able to set up bilateral co-operations with leading European and international groups at very early stages of the development of grids. For instance, the relationship between French groups and the group of Jack Dongarra in Knoxville, TN, have played an important role in the dissemination of the Network Enabled Servers (NES, for instance NetSolve) and the grid concepts. Also, very early contacts have been made between French groups and the Foster-Kesselman group, resulting in an early dissemination of Globus in France. More recently, an important effort has been put by CNRS and INRIA on contacting Asian countries in the area: Japan, Korea, and now China.