Research On Data Security And Network Performance Optimization Model Based On Column-stores

With the rapid development of network communication technology, Data created by the humans is growing exponentially. When big data is analyzed, only one or a few columns in a logical table is used. At the same time, large amounts of data need read. The column store database system was created in this application background. A large number of studies have suggested that read-mostly column store system has a significant advantage over row store system in the application environment with more query operation. In addition, data security also is an important question. How to maintain data security and keep the performance advantages of column storage system has become an important branch of research field in the column storage.This dissertation implements a systemic study on the issues of data encryption method, data security index, data distribution and reconstruction methods of system k-safe,Network model of optimizing system performance. The complete research is outlined below:Firstly, encryption scheme of cell granularity is researched. The cell data is combined with cell location, and then the traditional encryption technology is used to encrypt it.Among them, the cell location information is presented with a triple of table ID, column ID and row ID of column storage system tables. Finally, the encryption scheme is used to encrypt data. The security of the new encryption scheme which can resist replacement attack and statistical attack is proved by the concrete example analysis. On the basis of cell granularity encryption scheme, data security index method is research based on ciphertext. The cell data is combined with row location of cell, and then safe index is established based on ciphertext. Contrast analysis is carried on between ciphertext index and B+ tree index of plaintext. It is proved that any information related to plaintext data will not be leaked and the original index structure is keep.Secondly, According to the structural characteristics of column storage system, a light Romega encryption scheme is researched. Relative to row storage system, it has performance advantages in the read-mostly operation environment. Encrypting data willaffect system performance.To solve this contradiction, light encryption scheme is research based on multistage interconnection network. First of all, the structure characteristics of Omega is analyzed. Second, using its reversibility, a concrete encryption scheme is proposed. Finally, The security of the new encryption scheme is proved.Thirdly, the k-safe of column store system is studied. There are n nodes in a data grid.Of all the n nodes, the amount of data stored in each node is different. When data in k nodes is damaged, it is still able to reconstruct all the data by using the remaining data in n-k nodes. Based on the different application environment, two schemes of data distribution and data recovery are proposed by using lagrange interpolation polynomial and newton interpolation polynomial. Later, the correctness and feasibility of two schemes are analyzed by concrete examples.Finally, from research perspective of application environment, a queuing model is studied. First of all, the basic structure of traditional queuing theory and fluid queue is analyzed. And then the model assumption is put forward. According to the hypothesis, a new fluid queue model is proposed.