HIERARCHICAL DATA STORAGE FRAMEWORK BASED ON CLASS WEIGHTED MIGRATION DEEP Q NETWORK STRATEGY

In order to realize the efficiency and applicability of data management, a hierarchical data storage framework based on class weighted migration deep Q network strategy is proposed. The key of the proposed method was to improve its ability to conduct online dynamic data transmission between different storage solutions, and to make decisions based on data access mode and availability. Therefore, the class weighted migration deep Q network strategy was introduced to solve the data migration problem in the tiered storage system, while ignoring the source outliers, which effectively encouraged the transfer of positive knowledge and improved the effect of domain adaptation. A simulation software and a cloud framework were designed to test. The results show that the proposed method is efficient and adaptive.