Job ranking and scheduling in utility grids VOs

Abstract

In this work, we propose approaches to creation of a ranked jobs framework within a model of cycle scheduling in virtual organizations of utility Grids with the decoupling of users from resource providers. Two methods for job selection and scheduling are proposed and compared: the first one is based on the knapsack problem solution, while the second one introduces a heuristic parameter of a job and a computational resource set “compatibility”. Along with these methods we present experimental results demonstrating the efficiency of proposed approaches and compare them with random job selection.