DEISA – Cooperative Extreme Computing Across Europe
The European DEISA1 project is an infrastructure of infrastructures – eleven of the leading European High Performance Computing (HPC) centers in Europe interconnected by a 10 Gb/s high speed network – that devised an innovative strategy to enable the cooperative operation of existing national supercomputing infrastructures. This initiative led to the deployment and operation of a world class, persistent, production quality, distributed tera-scale supercomputing environment with continental scope enabling scientific discovery across a broad spectrum of sciences and technologies. The paper describes in detail the DEISA infrastructure and high speed network interconnect, its operation and management, as well as user support procedures. The second part explains the used Grid middleware and Grid security models as well as its integration into the local security policies. The third part of the paper gives a comprehensive overview about lessons learned. The paper closes with a summary and a vision on future DEISA project extension plans.
Unable to display preview. Download preview PDF.
- 1.The DEISA Project home page, Distributed European Infrastructure for Supercomputing Applications, see http://www.deisa.org
- 2.GÉANT2 is the seventh generation of pan-European research and education network, see http://www.geant2.net
- 3.IBM’s General Parallel File System, http://www-03.ibm.com/systems/clusters/software/gpfs.html
- 4.TeraGrid is an open scientific discovery infrastructure combining leadership class resources in the US, see http://www.teragrid.org/
- 5.GridFTP file transfer protocol, https://forge.gridforum.org/projects/gridftp-wgGoogle Scholar
- 6.Open Grid Services Architecture Data Access and Integration (OGSA-DAI) is a Globus project, see http://dev.globus.org/wiki/OGSA-DAI
- 7.Storage Resource Broker of SDSC, see http://www.sdsc.edu/srb/index.php/Main_Page
- 8.UNICORE (Uniform Interface to Computing Resources) is a ready-to-run Grid system including client and server software, see http://www.unicore.org/
- 9.Ganglia is a scalable distributed monitoring system for high-performance computing systems such as clusters and Grids, see http://ganglia.sourceforge.net
- 10.The Monitoring and Discovery Service (MDS) is the information services component of the Globus Toolkit, see http://www.globus.org/toolkit/docs/2.4/mds/
- 11.The Globus Toolkit, see http://www.globus.org
- 12.INCA - User level grid monitoring, see http://inca.sdsc.edu/