the paper introduces the parallel DBMS into the Hadoop framework and introduces the column-oriented storage technology into the copy and scheduling of Hadoop
to get efficient coding model.Experimental results show the scheme presented in the paper can improve the performance of MapReduce.