Briefs

Computerworld (04/05/04) Vol. 32, No. 14, P. 22; Thibodeau, Patrick

Large companies and public institutions are beginning to investigate data grids, which have garnered far less media attention than computing grids but work on the same principles. Just as computing grids allow resources to be viewed singularly and applied to a specific problem, data grids allow a single view of data stored in disparate physical locations. Data grids use metadata frameworks and middleware to enable the pooling of data resources, irregardless of the operating systems or formats used in the databases. Data and computing grids can operate together. So far, most research institutions have adopted data grids, which are best applied to organizations with large and scattered data repositories. Standards need to be established and widely adopted among vendors for data grid applications to be most effective, says Pfizer research information architecture director Paul Lewis. While current data grid products are being deployed within organizations, the most advantageous deployments will be across companies and supply chains, say experts. Globus Alliance grid standards co-director Ian Foster compares the protocols that enable remote data access to Internet Protocol. He says that fixed-file-based data is easy to link up the grid, but more work needs to be done on relational databases and XML databases. European research groups are linking up their data repositories in the DataGrid project, and are gathering resources for an even larger endeavor called Enabling Grids for E-science in Europe (EGEE). Middleware and security are outstanding issues, says DataGrid project head Frabrizio Gagliardi. Eventually, data grids may reduce hardware costs since enterprises can use more heterogeneous vendor offerings.

Here is the rest of the story.