An Improved Concurrent Programming Architectural Model Based on Cloud Computing

MapReduce, as one of the main concurrent programming models based on cloud computing, has become the research hot spot of information technology. Aiming at the development of MapReduce application program of high quality and efficiency, the working mechanism based on the Hadoop MapReduce model is analyzed in this chapter and MapReduce concurrent workflow is elaborated at the level of development class library, including task creation, job initialization, task initialization, communication between task and job. In addition, in order to solve the problem of Reduce input imbalance, a universal Map-Balance-Reduce improved model is proposed in this chapter. The balance layer embedded an adaptive splitting algorithm is added to the MapRduce model before reduce targeted at Reduce’s defect of input imbalance, and its function is to guarantee the balanced Reduce input with the semanteme unchanged. The simulation indicates that the unbalanced degree of the improved MBR is obviously lower than that of MR; finally, some improvement prospects of the open source MapReduce model are discussed.