Cloud computing jobs are rapidly increasing with the expansion of the scale of cloud computing. In order to make the cloud computing service more stable and efficient, the fast and balanced job allocation algorithm has become a focus in the study of cloud computing. Based on the characteristics of cloud computing, we analysis the nodes’ computing power. Group the node under the principle of Quicksort based on its computing power. Task was assigned to the corresponding node group according to its size under the order of the task. The optimal computing node was chose by greedy algorithm. The result shows that this method is far better than Round Robin algorithm and has faster allocation time than total Greedy Algorithm, which was proved by CloudSim.