More than 90 members of the Open Science Grid (OSG) consortium met recently in Boston to plan for the deployment of a production Grid infrastructure for science in the US.
The OSG will provide a coherent environment for the peta-scale data management and computational needs of diverse scientific teams, from those with a single member to those with thousands of members. The goals of the OSG are to integrate existing and emerging processing and storage facilities, to provide a production-quality scalable supporting Grid infrastructure and to enable effective throughput of a broad range of scientific analyses.
Participants in the consortium come from the scientific communities and include Grid technologists and institutions and projects with experience in Grid technologies, deployment and applications. The OSG is an evolution of Grid3, which has been supporting a general load of around 800 simultaneous jobs for US-ATLAS and US-CMS 2004 data challenges, and up to 1200 jobs including astrophysics, gravitational physics, bioinformatics and computer-science applications.In particular, the opportunistic computing benefits of Grid3 have already stimulated the US-ATLAS and US-CMS teams to share their resources and accommodate each others' priorities and schedules. The first deployment of OSG is planned for 2005 and the infrastructure will be based on the evolution of the Virtual Data Toolkit, which is already the common basis for Grids for particle physics. Grid3 will be extended to include resources and applications from a broader mix of applications from other sciences, as well as those from particle-physics experiments already running or in production (e.g. CDF, D0, BaBar, STAR etc).
The OSG roadmap was proposed in a white paper in the summer of 2003. The first workshop in January 2004 served to demonstrate a broad level of community support for the direction the map outlines. This was followed by the formation of several technical working groups, increased alignment of related ongoing activities in the US with the longer term direction, and further discussions with the Department of Energy and National Science Foundation sponsors.
An important focus of the OSG is to federate and overlap with other Grid projects - multi-disciplinary computing facilities, university and regional campus Grids, large high-energy and nuclear physics facilities (at Fermilab, Brookhaven, LBNL and SLAC), and national infrastructures such as Teragrid. In particular, all the working groups of the consortium collaborate with their peers in the LCG and EGEE projects towards a coherent global infrastructure for physics at the LHC.
The Boston workshop provided the opportunity to take the next steps in solidifying the technical programme, increasing the involvement of a broad range of participants and in formulating the organization of the OSG consortium itself.
• For further information, see http:// opensciencegrid.org/events/meetings/boston0904/index.html
Author:
Compiled by Hannelore Hämmerle and Nicle Crémel