Research on distributed competition big data hierarchical storage method based on Ceph architecture

Abstract

In the face of intensive data access requests, the current big data storage methods have the problem of poor concurrent read-write performance. Based on Ceph architecture, a distributed big data hierarchical storage method is designed to realize the robust storage service of data files. According to the frequency and popularity of the access object, the cold data with low frequency is eliminated, and the access cache pool or back-end storage pool is selected. Standardize the I/O performance, load and capacity values in the OSD set, calculate the comprehensive weight, and determine the priority of node access. The entire Ceph architecture balances the data distribution. When highly concurrent data access operations occur, a random number is selected to execute the read service of the storage node to ensure the interconnection communication in different clusters. The test results show that the write and read time of the hierarchical storage method based on Ceph architecture is greatly reduced, and good results are achieved in terms of cache hit rate, which can effectively manage and store competitive big data.