GridCoord DoW
Yüklə 0,53 Mb.
|
- Bu səhifədəki naviqasiya:
- Network technologies
- Large scale experimental platforms
- Software infrastructure
- Exploitation of Grids in Applications
- Statistics
- Methodology
- National projects
- European projects
InfrastructureThe second area of interest in Grid research is related to the Hardware/Software infrastructure that is deployed as a Grid. This comprises high-bandwidth network technologies, deployment and management of large scale platforms, and the service infrastructure that must be deployed for a collection of resources to become a Grid. Network technologiesSeveral national projects: e-Toile and VTHD++ in France, VIOLA in Germany, Grid.it in Italy and GigaPort in the Netherlands, explicitly target next generation network technologies as an important research area. They focus on next generation optical technology (wavelength division multiplexing), integrated with quality of service management and monitoring. Large scale experimental platformsNational projects aimed at building large scale experimental grid platforms (Grid'5000 in France, Grid over the GARR research network in Italy, Clusterix in Poland, ClusterGrid and the Hungarian SuperComputing Grid in Hungary), catalyse the Grid research, providing the research community with a concrete experimental platform for large scale testbeds. As reported in the Annex 2 of the French contribution, we notice the difference between production grids (Clusterix, ClusterGrid, Grid over the GARR research network), used as number crunching tools with quality of service requirements, and research grids (Grid'5000, Hungarian SuperComputing Grid), more suitable to run experiments on the concepts of grid computing, being flexible, with fast answer times, and open. Software infrastructureA software infrastructure is needed to build dependable, consistent and pervasive services that are needed for a collection of heterogeneous resources to become a Grid. Developing a full-featured infrastructure requires a huge effort, and is currently carried out in Europe with international projects like EGEE (Enabling Grids for E-sciencE), CROSSGRID (interactive applications on Grids) and ONTOGRID (knowledge intensive applications on Grids). Given that basic infrastructures for Grids reach a certain stability supplementary efforts are needed and spent to enrich the set of basic services by adding higher-level abstractions:
Exploitation of Grids in ApplicationsGrid applications motivate the existence of Grids and the efforts to apply Grid research are plentiful. They can be developed as demonstrators in projects developing software development tools (Grid.it, Italy) or as real-life test cases in large scale Grid infrastructures (Grid'5000, France, and SuperClusterGrid, Hungary). Some applications or application classes, for their scientific importance, are themselves the target of the research, and developed in specific projects:
StatisticsGridCoord members provided a considerable amount of information about their national research projects and their partecipation to european projects. This information is qualitatively summarized in the previous section. This section tries to draw quantitative considerations. MethodologyBefore starting with the analisys of figures 1-6, it is worth saying something about the metodology employed in producing these figures. The first step was the detailed analisys of data received from the gridcoord partners. This was not an easy task, in particular related to the quantitative data (e.g. funds). All data relative to national projects are taken from gridcoord member contributions. Unfortunately in some contributions, this data lacks stable/reliable quantitave information on fundings. This impairs our ability to weight projects w.r.t. their economic relevance. Consequently, to analyse national projects, we applied the categorization introduced in the previous section, and portrayed unweighted project frequencies. Concerning international projects, all data are taken from the official CORDIS website (http://www.cordis.lu/ist/grids/index.html). For each project we could collect: aggregate cost and funding, duration and partners, while actual fund allocation to the partners was not available. Our analysis considered the projects reported in the following table, divided by the FWP they belong to.
National projectsFigure 1 shows how the national projects are classified by research area as application driven, infrastructure driven or driven by software technology. The picture shows that most national projects focus on application development (20) and infrastructure deployment (16), while software technology (6) is only marginally represented. Figure 2 shows the distribution of research areas v.s. the considered countries. The figure shows that in France the research is mainly conducted in the infrastructure area, boosted by the Grid’5000 project, and also Hungary is investing in Infrastructures, while Germany and U.K. focus mainly on exploitation of grids in applications. Italy, The Nederlands and Poland balance quite well their investments/efforts between Infrastructure, Applications and Research in Software Technologies, while Spain balances Infrastructure and Applications, neglecting Software Technologies. These countries represent well in the small the situation at the european level, shown in Figure 1. Figure 3 shows the main topics and their relevance in application development projects: Astrophysics (22%) and Bioinformatics/Biomedical (22%) are the leading topics, followed by Computational sciences and Engineering (9%), Data Intensive applications (9%) and High Energy Physics (9%). The research topics of Infrastructure oriented projects are summarized in Figure 4: it is worth noting that projects aimed at the deployment of large scale Grids (50%) balance the ones aimed at research (in software infrastructure (25%) and network technology (25%)). These research areas are balanced, while considering Grid deployment, we note that production platforms (37%) overwhelm research ones (13%). Figure 2 Figure 3 Figure 4 European projectsStarting from the links to international projects provided in the member contributions, and exploiting project data from previous researches (http://www.cordis.lu/ist/grids/index.html), some statistics have been drawn, to show how european funding is allocated. We collected public data for a considerable part of the mentioned projects, essentially the ones belonging to FP5 or FP6 (see Figure 5). For each project, we considered aggregate cost and funding, duration and number of partners. Since we do not have access to information regarding actual fund allocation to the partners (typically this information is reported in consortium agreement and is not publicly available), we assumed for each partecipant of a project the same weight. Project weights were computed by dividing the total funding of each project by the project duration and the number of partners. 
Figure 5 Figure 6 shows that the GridCoord members receive more or less half of the funds of the selected projects, and that among these, Germany (10%) and U.K. (9%) are the most active, followed by Italy (7%), France (5%), Poland (5%) and Spain (5%). Note that Sweden did not provide data, so its small share is only due to partecipations to projects mentioned by other, and can therefore be inaccurate. Yüklə 0,53 Mb. Dostları ilə paylaş:
|
