云计算的数据存储模型研究及应用

发布时间：2018-05-12 18:24

本文选题：CP-ABE加密算法 + Map/Reduce模型　；参考：《湖南大学》2013年硕士论文

【摘要】：随着信息技术的广泛应用和快速发展，云计算作为一种新兴的商业计算模型日益受到人们的广泛关注。然而，云计算数据存储模型的安全性缺乏可靠性。鉴于此，本文主要对云计算用户数据存储模型进行了研究分析，并通过对相关的机器学习和构建CP-ABE算法模型，改进Map/Reduce算法模型，提升访问速度。本文在前人研究的基础上，主要从事了以下几方面的工作：首先，针对传统对称加密算法存在诸如自身安全性不高、加密和解密采用同一密钥而带来密钥传播安全等固有问题、以及非对称加密体系运算量大不适用于大数据处理的弊端，通过研究密文属性的数据存储算法CP-ABE，该算法利用了单调访问控制树，将服务器端的相关属性集合在内部节点和叶子结点的访问控制树中，从而保证了云计算系统的安全可靠，并且该算法相比于传统的对称和非对称加密算法具有密钥服务器端集中管理、使密钥相对于用户透明，并更易于管理。在云存储系统中，基于CP-ABE算法的访问控制模型可以由加密方控制主动权，只有满足密文要求的用户才能获得解密权限。该机制降低了权限管理的复杂度以及访问控制信息的存储空间；其次，由于Map/Reduce模型在分配输入文件时没有考虑集群中大量异构节点的计算性能，导致运行map任务时网络数据传送时间增加，本课题针对该问题提出一种云计算环境下的改进型Map/Reduce模型，以降低分类过程中的模糊粒度。针对云计算环境中不同的节点在集群中的计算性能不同等特点，通过设计2次改进型索引分类存储模型和算法，在传统的1次Map/Reduce结果中建立反向索引表，并在2次Map/Reduce计算中对索引词进行权重计算分类，，这样索引数据经过2次分类后，不再是简单的反向索引表，而是具有权重阈限的分类反向索引表，从而提高数据存储能力和计算速度；最后，结合学校实验条件状况，搭建hadoop平台模拟大型异构云环境，验证文中提出的改进Map/Reduce算法模型的有效性和效率。
[Abstract]:With the wide application and rapid development of information technology, cloud computing as a new business computing model has attracted more and more attention. However, the security of cloud computing data storage model lacks reliability. In view of this, this paper mainly studies and analyzes the cloud computing user data storage model, and through the related machine learning and the construction CP-ABE algorithm model, improves the Map/Reduce algorithm model, enhances the access speed. On the basis of previous studies, this paper mainly engaged in the following aspects of work: First of all, the traditional symmetric encryption algorithm has some inherent problems, such as the low security of itself, the security of key propagation caused by the use of the same key in encryption and decryption, and the disadvantages of asymmetric encryption system, which is not suitable for big data processing. By studying CP-ABE, a data storage algorithm for ciphertext attributes, the algorithm utilizes monotone access control tree, and sets the related attributes of server in the access control tree of internal nodes and leaf nodes, thus ensuring the security and reliability of cloud computing systems. Compared with the traditional symmetric and asymmetric encryption algorithm, this algorithm has the key server side centralized management, which makes the key more transparent than the user, and easier to manage. In cloud storage system, the access control model based on CP-ABE algorithm can be controlled by the encryptor. Only the user who meets the requirements of ciphertext can obtain decryption permission. This mechanism reduces the complexity of privilege management and the storage space of access control information. Secondly, because the Map/Reduce model does not consider the computing performance of a large number of heterogeneous nodes in the cluster when allocating input files, the network data transfer time increases when running the map task. In this paper, an improved Map/Reduce model in cloud computing environment is proposed to reduce the fuzzy granularity in the classification process. In view of the different computing performance of different nodes in cloud computing environment, by designing two times improved index classification storage model and algorithm, the reverse index table is established in the traditional one-time Map/Reduce result. In the second Map/Reduce calculation, the index words are classified by weight calculation. After two times of classification, the index data is no longer a simple reverse index table, but a classified reverse index table with weight threshold. In order to improve the data storage capacity and computing speed; Finally, the hadoop platform is built to simulate the large heterogeneous cloud environment, and the effectiveness and efficiency of the improved Map/Reduce algorithm model proposed in this paper are verified.
【学位授予单位】：湖南大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP333

【参考文献】