Grid Computing
The SCOOP program will achieve its goals for environmental prediction by creating the SCOOP Grid, which will leverage other SURA IT and Grid initiatives. In recent years, Grid computing has evolved as a distributed system that is focused on resource and data sharing to form virtual communities possibly across geographical boundaries. Thus Grids comprise a network of shared resources of all kinds, including computing power, databases, instruments, applications, storage, and more. The challenges to creating a Grid are both technical and social. Grid middleware, such as the Globus toolkit (www.globus.org) are commonly associated with Grid applications, but they provide only part of the solution. The ability to apply these technologies effectively relies upon the community agreements for information exchange. In conjunction with other SURA activities such as SURAGrid, the SCOOP partners are leveraging Grid technologies to provide the ocean modeling community access to distributed resources and data.
SCOOP Grid activities widely uses Grid and web service technologies and existing standards. The SCOOP activities focus on both computational, through distributed resource sharing, as well as data aspects of Grid environments, through Open Geospatial Consortium web services.
SCOOP partners are actively engaged in the Grid computing research area focused on scheduling, fault tolerance, virtualization, application toolkits, etc. Several simulation models developed by SCOOP partners are slated to benefit from the availability of computational Grids, but ensuring that they seamlessly run on the heterogeneous software environments typical of Grid resources is challenging. SCOOP partners are also leveraging novel technologies that use virtualization software as a complement to traditional Grid middleware to facilitate the porting of models and foster a gradual transition of various applications into future Grid environments.
|