Date Added: May 2010
The scale and complexity of Grid systems constantly increases. The largest currently Grid infrastructures, such as the LHC Computing Grid, consist of hundreds of resource centers located world-wide providing tens of thousands of CPUs. It is expected that future Grids will achieve an even greater scale. As the number of sites, nodes, and users in the Grid grows, efficient job scheduling becomes increasingly difficult and traditional approaches to job scheduling gradually become inadequate. Centralized schedulers, often used in single clusters, do not scale and introduce policy problems since they require a single authority to coordinate the resources of the entire Grid.