基于块级的分级存储数据特征模型及其应用研究

Research on the Tier Model and Application of Extent-Based Data Feature

随着信息爆炸式地增长,以高效、合理使用存储资源为目的的数据分级存储管理已然成为存储领域的研究热点.但是,数据分级存储管理过程中存在热点数据定位不准确的问题,主要原因是对基于块级的数据特征研究不够充分.针对上述问题,提出一种基于扩展块的数据特征模型EDM.EDM在基于扩展块级粒度的基础上建立分级数据管理的特征模型,通过对被管理数据对象的特征进行统计分析,自动进行数据管理决策,将冷/热数据迁移至对应层级以便能够高效合理地使用存储资源.最后,设计实现了基于EDM模型的原型系统,通过对原型系统的测试表明EDM模型具备可行性,能够有效管理存储资源.

With the explosive growth of information,hierarchical storage management(HSM)has become a hot research field,and the purpose of HSM is to use the storage resource effectively. However,since insufficiently research on extent-based data feature,hot data can't be found accurately.In order to deal with the problems,an extent-based data feature model,EDM,is proposed,which is a model of hierarchical data management based on block-level granularity.By statistically analyzing the characteristics of data objects,EDM can make the management decisions, and migrate the cold or hot data to the corresponding level to be able to efficiently and reasonably use the storage resources.The prototype system test shows that EDM model is feasibility and can effectively manage storage resources.