CO-ALLOCATION IN GRID COMPUTING USING RESOURCES OFFERS AND ADVANCE RESERVATION PLANNING

Abstract

Computational grids have the potential for solving large-scale scientific problems using heterogeneous and geographically distributed resources. However, a number of major technical hurdles must overcome before this potential can be realized. One problem that is critical to effective utilization of computational grids and gives a certain Quality of Service (QoS) for grid users is the efficient co-allocation of jobs. The advance reservation technique has been widely applied in many grid systems to provide QoS, however, it will result in low resource utilization rate and high rejection rate when the reservation rate is high. This work addresses those problems by describing and evaluating a grid resources co-allocation algorithm using resources providers offers and planning the advance reservations. In our algorithm, a Metascheduler performs job scheduling based on resources offers and use advance reservation planning mechanism to reserves the best offers. Offers act as a mechanism in which resource providers expose their interest in executing an entire job or only part of it. The Metascheduler selects computational resources based on best offers provided by the resources; Meta-schedulers can distribute a job among various clusters that are usually heterogeneous in order to speed up the job execution. The main aims of our algorithm is to minimize the total time to execute all jobs (Makespen), minimize the waiting time in the global queue, maximize the resources utilization rate and balance the load among the resources. The proposed algorithm has been compared with other scheduling schemes such as First Come First Served (FCFS), easy backfilling (EBF), Fit Processor First Served (FPFS) and a simple co-allocation algorithm without offers support (SCOAL). The proposed algorithm has been verified through an extension of GridSim simulation toolkit and the simulation results confirm that the proposed algorithm allow us to achieve our goals by minimizing the Makespan and the waiting time, maximizing the resources utilization rate and load the balance among the resources.