Towards the Design of an Active Grid
ICCS '02 Proceedings of the International Conference on Computational Science-Part II
Active Networking Support for the Grid
IWAN '01 Proceedings of the IFIP-TC6 Third International Working Conference on Active Networks
Hi-index | 0.00 |
The Master-Worker distributed computing paradigm has proven to be a very effective means for harnessing the power of computational grids. At any given time, the master of the application controls a collection of CPUs that has been allocated to the application by the resource manager of the grid. Effective management of this dynamic "workforce" of CPUs holds the key to ability of the application to meet its computational objectives. Like in similar real-life situations, the master has to decide on a target size and composition for the workforce, a recruiting strategy and a dismissal policy. It has to decide on who does what and how to deal with workers that do not complete their assigned task on time.