A Grid Simulation Infrastructure Supporting Advance Reservation

Abstract

Advance Reservation (AR) for global grids becomes an important research area as it allows users to gain concurrent access for their applications to be executed in parallel, and guarantees the availability of resources at specified future times. Evaluating various AR scenarios can not feasibly be carried out on a real grid environment due to its dynamic nature. Therefore, we extend a GridSim simulation package to support AR for repeatable and controlled evaluations. This paper discusses the design and implementation of AR within GridSim, together with the effects of AR from users' and resources' point of view in the experiments.