GDCS resource deployment strategy of conflicting & sharing

Reasonable resource deployment is an important guarantee for grid effective service. From industry application perspective, the resources is constrained by property, so as to GDCS have to deal with conflicts resulting from resource sharing. Based on the hierarchical task scheduling model, the incomplete resource deployment strategy was proposed in this paper. Firstly, queuing theory is used to solve the single resource allocation of multitasks. With known transmission rate and execution rate, and aiming for the lowest cost of communication and execution, multitasks are sorted, and resources exploring right are allocated according to task priority. Secondly, under conditions of completely independent or incompletely independence, the resource deployment heuristic algorithm are proposed to solve multi-resources share and conflict caused by multitasks: the serial transmission of task assignment is implemented by system scheduler, and then resources is deployed by region scheduler in parallel. The results proved that the burden of system communication is lessened, and furthermore, the centralized work pressure of system scheduler is lightened. However, this algorithm can obtain nearly optimization solution instead of optimum solution. Finally, the future research directions of this technique are also discussed.