Grid resource management is regarded as an important component of a grid computing system. The resources in the grid are geographically distributed, heterogeneous in nature, owned by different individuals/ organizations each having their own resource management policies and different access-and-cost models. Based on Abstract Owner model, this article proposes a new resource storage model called resource pool which couples geographically distributed resource together. Meanwhile, this paper uses queuing and global optimization theory as tools to emphasize on quantitative analysis for resource pool capacity. Consequently, some new analytic formula for practical computation is obtained. Furthermore, an algorithm for analytic formulas is introduced into global optimization to evaluate the performance.