基于模糊逻辑的数据分级存储模型研究

Research on Fuzzy Logic-based Model of Tiered Storage

数据分级存储是智能数据管理的重要途径,利用分级存储能够有效地平衡存储资源与不同数据之间的存取关系,最大程度地提高存储系统的整体性能。但是,在数据分级过程中对于数据特征的发现以及热点数据的判断一直是数据分级存储的瓶颈。提出一种基于模糊逻辑的数据分级存储特征模型FLM,该模型将反映数据冷热程度的关键特征作为输入量,利用模糊逻辑对热度特征量进行推理获得输出量,从而平滑热点数据与非热点数据的边界,避免尖锐边界问题,以利于数据迁移的平顺性,降低数据管理中出现的抖动问题。

Iered storage is the important way for Intelligent Data Management. It can effectively balance access relationships between storage resources and variety data, and maximize the overall performance of the storage system. However, the judgment of hotspot data is a bottleneck for tiered storage in process of data classification. Proposing a Fuzzy Logic-based Model of tiered storage （FLM） which uses the data characteristics as the input variable that reflect hot and cold level of data. Then, FLM uses the fuzzy logic to analyze data characteristics for getting output variable, and it can smooth the boundary of non-hotspot and hotspot. Experimental model analysis indicates that FLM can make the smoothness of data migration and reduce shake problems in data management.