Considering the confidentiality and integrity of big data in cloud storage, a MapReduce-based AES parallelization scheme is designed by using MapReduce framework of the open source Hadoop in this paper. The scheme takes full advantage of MapReduce and modern cryptography technologies to parallelize AES encryption and decryption process, in the way of data decomposition, which speeds up the efficiency in the implementation of encryption and decryption. Meanwhile, mix plaintext, separate storage and other technical means are taken into account in this scheme to ensure the confidentiality and security of the key and the ciphertext. By analyzing the performance, it is proved that the time consumption of new scheme is significantly reduced comparing with the traditional method.