云系统垂直分层布隆过滤漂移数据删除技术

发布时间：2019-02-22 18:08

【摘要】：云系统中的漂移数据作为大量冗余数据的一种,对其有效及时的删除能保证云存储系统的稳定与运行。传统的云系统漂移数据删除技术采用全文件静态分块策略,由于操作粒度较小,对漂移数据的删除率不高。提出一种基于垂直分层布隆过滤的云系统漂移数据删除算法,设计基于客户端/服务器的云存储系统漂移数据删除机制体系构架,采用布隆过滤技术进行数据源端处理,引入热度垂直索引热度来衡量数据块边界偏移所造成用户的访问集中热点损失,文件根据内容划分成可变长度的数据块,进行垂直分层,得到备份集中数据对象的粒度,采用奇异值分解的方式对漂移数据序列的细节信号展示,根据矩阵奇异值分解矢量的唯一性,避免一些数据块边界偏移造成的误删和漏删。仿真实验表明采用该算法进行云系统的漂移数据删除,性能较好,执行效率和精度优越于传统算法。
[Abstract]:As a kind of redundant data, the drift data in cloud system can be deleted effectively and timely to ensure the stability and operation of cloud storage system. The traditional cloud drift data deletion technique adopts the static block strategy of full file. Because of the small granularity of operation, the deletion rate of drift data is not high. This paper proposes a cloud drift data deletion algorithm based on vertical tiered Bulon filtering, designs a client / server based cloud storage system drift data deletion mechanism architecture, and uses Bulon filtering technology to process the data source at the end of the system, and then designs the cloud storage system drift data removal mechanism architecture based on client / server. The hot vertical index heat is introduced to measure the hot spot loss of user's access set caused by block boundary offset. The file is divided into variable length data blocks according to the contents, then vertical tiering is carried out, and the granularity of data objects in the backup set is obtained. The detail signal of drift data sequence is displayed by singular value decomposition (SVD). According to the uniqueness of matrix singular value decomposition vector, some data block boundary offset errors and omissions are avoided. Simulation results show that the proposed algorithm is more efficient and accurate than the traditional algorithm in removing drift data from cloud system.
【作者单位】：吕梁学院计算机科学与技术系;
【分类号】：TP333

【参考文献】