ePubs

The open archive for STFC research publications

Full Record Details

Persistent URL http://purl.org/net/epubs/work/30104
Record Status Checked
Record Id 30104
Title Data Virtualisation in the NERC DataGrid
Contributors
Abstract Research in the earth sciences requires access to large and complex datasets. These data include in-situ and remote-sensed observations, and a variety of model output. Storage methods are as diverse as the data types and include numerous file formats and relational database systems. In practice, considerable effort is expended in data handling. Processing software choice is constrained by file format compatibility, and data object representations are coerced onto physical storage artefacts. A key goal of Grid technologies is to facilitate virtualisation of resources. Essential semantic behaviour and content is abstracted from low-level implementation. Virtualisation may be applied to data, storage and computational resources. This paper presents details of the NERC DataGrid data model, and the virtualisation of earth science data it enables. The model is based upon nested hierarchies of multidimensional arrays. Standard profiles of the model are defined for important data types like 4-D gridded meteorological forecast data or oceanographic cruise measurements, for instance. Rich georeferencing information is incorporated. An XML schema provides the mechanism for mapping physical storage artifacts onto the data objects provided by the model.
Organisation CCLRC , ESC , ESC-DMG
Keywords earth science , virtualisation , data model , grid , TC211 , NDG
Funding Information
Related Research Object(s):
Licence Information:
Language English (EN)
Type Details URI(s) Local file(s) Year
Presentation Presented at All Hands 2003, Nottingham, England, 2 Sept 2003 - 4 Sept 2004. ndgdatamodel.pdf 2003