In the actual grid environment, there are many kinds of task and resource which are different from each other, so the grid scheduling is very difficult. This paper proposes a scheduling algorithm for static batch task based on graph division and resources clustering. The algorithm uses VDAG map to describe the entire task collection and schedules them together. In order to take full advantage of the parallel grid resource, the algorithm takes a method to divide the DAG graph in horizontal and vertical direction to get the job groups relatively independent from each other; at the same time, extracts the resources attributes to build the virtual cluster to reduce the resources search time spent in the scheduling process. The algorithm obtains better scheduling results.