摘要
提出一种基于内容感知存储系统的自动分级存储模型.该模型从文件中提取内容元数据作为分级特征信息,并以领域本体表示分级存储规则的语义.采用一种基于特征词权重的算法计算分级特征对领域本体的支持度,并通过分级判断规则得出文件的存储级别.算法通过引入修正权重系数提高了基于支持度的分级准确率.实验结果表明,这种模型对于携带特征信息量较多的数据能实现较高的准确率,对系统写入性能的影响随着系统平均数据大小的增加而降低.
This paper proposes a auto-tiered storage model for designating storage levels for files in a content aware storage system.The model retrieves content metadata from files as the characteristic information and build up a domain ontology to represent the semantics of tiering rules.A weight-based algorithm is used to calculate the support degree of characteristic information to domain ontology,and the storage level is decided with the support degree and tiering judgment rules.By introducing a correction factor,the algorithm can efficiently improve the tiering accuracy.The experimental results indicate that this model can achieve higher precision with data which carries more information with it,and the impact on writing throughput will decrease with the increasing of average size of data.
出处
《小型微型计算机系统》
CSCD
北大核心
2011年第6期1151-1156,共6页
Journal of Chinese Computer Systems
基金
部委基金"基于服务定制的智能存储系统研究"项目资助
国家自然科学基金项目(606730001)资助
国家"九七三"重点基础研究发展计划项目(2004CB318203)资助
关键词
内容感知存储
自动分级
领域本体
权重修正系数
content aware storage
automatically tiering
domain ontology
weight correction factor