Data Resource Discovery in a Computational Grid

Abstract

A computational grid focuses on the effective sharing of computing resources instead of data resources over the grid. However, distributed or parallel scientific and engineering applications often require wide access to large amounts of data. The demands to serve data are often as important as computational resources. Therefore, our focus is on the discovery of data resources leveraging on existing computational grid management technologies. In this paper, we proposed a data resource discovery mechanism to allow the Sun Grid Engine to retrieve, index and monitor the metadata from Oracle 10g. In the implementation, we analyze the organization of metadata information in Oracle 10g – Oracle’s data grid solution, and the resource discovery mechanism in Sun Grid Engine, one of the widely used computational grid middleware. The proposed mechanism provides a unique method to share data resources in the same way as computational resources. This allows the computational grid to be better integrated with the data grid.