中文摘要：

为了达到动态地根据用户的需求来实现数据的预取和缓冲的目的,提出了基于聚类的动态副本创建策略（DRCC）.该策略根据用户的访问历史来抽取访问数据的特征,判断这些文件之间的相关性,并通过聚类操作形成用户的相关文件类集合,然后根据该集合进行数据文件副本的创建.实验结果表明,DRCC策略可以有效地识别出客户节点的访问需求,并据此进行文件的复制,从而缩短了客户应用的访问时间,提高了访问效率,同时该策略还具有较强的应用适应性.

英文摘要：

For the aim of prefetching and buffering data,a strategy called strategy of dynamic replication creation based on clustering（DRCC） is proposed.It abstracts the characteristic of access data according to the access history of users.And then,judges the correlation among the files.And then,through clustering operation creates the correlative file class sets relating to the access requirements of users.It also produces the data files replica on the basis of those sets.The simulation results show that DRCC has the ability of identifying the access requirements of client nodes effectively.On the other hand,the replication of files based on it reduces the access time of client nodes and improves the efficiency of system.In the mean time,DRCC has greater applied adaptability.