Abstract
Abstract： The grid system consists of many different distributed resources, which can supply high-speed computing efficiency to deal with several mass and complex workloads. Thus, the Grid computing is a powerful and fundamental technology in the future. In Grid systems, a resource management system (RMS) would assign many subtasks to adequate resources for sake of better service cost (time or reliability). Nowadays, lots of methodologies (such as the SDP, IE, and UGFM et al.) can exactly calculate the service cost (time or reliability) of a grid system, but those have high time complexity when service require was evaluated in a large-scale system. Therefore, we applied Cellular automata Monte-Carlo simulation to estimate approximate service reliability of a large-scale grid system. In this paper we computed total cost by system reliability and maximum time. Finally, we adopted Genetic Algorithm to search the best assignment combination of subtasks and resource for optimal service total cost.