The GridSim firstly summarizes and defines data grid models and the process of job scheduling, simultaneously analyzes the time and cost of job execution in the data grid. The job scheduling strategy influences the QoS of the data grid immediately and then proposes a design proposal of the job scheduling simulator of data grid based on a grid simulator named GridSim and introduces the architecture, process and key technologies of the job scheduling simulator. By using this simulator we can determine the resource failure so that the performance of the data grid can be analysed. This involves implementation of job scheduling in a grid for handling resource failure using GridSim simulator. Moreover, the resources may suffer failures, and all of this obviously would affect the performance received by the users. In this paper we present an extension to one of the most popular grid simulators GridSim to support variable resource availability. Through which the overall performance can be determined. Finally, it proves that this scheduling simulator can satisfy the need of research on the data grid optimization in Grid Computing.