Load balancing in distributed simulations on the grid

Multiprocessors, networks of workstations and cluster computers have well served the computational needs of the simulation community by supporting the distributed simulations of larger models. However, the application of simulation to more complex problems calls for ever more computational power. The emergence of the grid provides an unrivalled opportunity to address this problem and thus facilitate advances in the modelling and the analysis of large-scale systems by harnessing the power of many computers. The aim of this paper is to present a framework for the execution of optimistic distributed discrete event simulations on the grid. A generic architecture is described and a load-balancing scheme is introduced in the context of a grid-aware TWarp simulation kernel.