基于模式识别的分布式存储信息一致性控制

Consistency Control of Distributed Storage Information Based on Pattern Recognition

为降低主副本数据更新延时、减少存储开销大的问题,解决分布式存储信息出现不一致的情况,提出基于模式识别的分布式存储信息一致性控制方法。通过组建中心控制端、传输控制端以及移动端构成分布式信息存储机制,了解其具有主副本数据延时更新的问题;将数据全相关一致性更新技术与模式识别法相结合,选取最优特征子集,将添加、修改、删除三种操作下形成的数据集定义为结构序列,以状态向量的记录作为数据一致性主要参考,分析对比两个节点数据库的状态向量,完成分布式存储信息一致性控制。通过仿真表明,所提方法可以实时更新主副本数据,节省存储开销。

Traditionally, the distributed methods have many defects, such as high update delay of primary replica data, large storage overhead and inconsistent distributed storage information. Therefore, this paper proposes a control method of distributed storage information consistency based on pattern recognition. First of all, through the combination of central control end, transmission control end and mobile end, the distributed information storage mechanism was constructed in order to understand that it had the primary replica data delay update. Secondly, the optimal feature subset was selected based on the update technology and pattern recognition method of data full correlation consistency. Then, the data set formed by adding, modifying and deleting operations was defined as structure sequence. Moreover, the record of state vector was used as the reference of data consistency. The state vector of two node databases was analyzed and compared in detail. Finally, the consistency control of distributed storage information was completed. Simulation results show that this method can update the primary replica data in real time and save the storage cost.