Increasing the throughput of available resources using management tools based on Grid technologies

The aim of this work is to build a dynamic virtual high performance platform made up of the heterogeneous individual machines available in a given instant of time in a work-center. In order to achieve this objective a low-intrusive monitor and scheduler system layered on top of an already existing Grid technology is proposed. The monitoring tool indicates available/unavailable nodes to be added/removed to/from the virtual platform. The scheduler manages the distribution of jobs among the available resources. The final goal is to increase the computational capacity of the work-center in a way that is both simple and costless for the users. Results obtained using the proposed tools to run two different applications, an image modeling application and an air quality model simulation, prove the feasibility and efficiency of our proposal.