With the explosively increase of social networking, bioinformatics network and other relational datas. It is urgent to improve the processing capability on large scale graphs with billions of vertices. The traditional graph algorithms can no longer meet the rapidly growing demand for the calculate dependence of the memory. To find the Weakly Connected Components of a graph, this paper presented a fast, scalable and iterable coloring algorithm CR and also established the MapReduce model. At last, we give out a contrast experiment between the open-source data mining toolbox XRIME and CR on four sets of communication data, which is provided by the Social Network Analysis Laboratory in Stanford University.