基于分布式FTP的二进制大对象存取算法  

Binary Large Object Storage Algorithm Based on Distributed FTP

作　　者：谢华成[1] 谢蕾[1] 刘道华[2] 

机构地区：[1]信阳师范学院网络信息与计算中心,河南信阳464000 [2]信阳师范学院计算机与信息技术学院,河南信阳464000

出　　处：《信阳师范学院学报（自然科学版）》2012年第3期391-394,共4页Journal of Xinyang Normal University(Natural Science Edition)

基　　金：河南省科技攻关项目(102102210175;112300410234);河南省教育厅科技攻关项目(2011A520038)

摘　　要：为提升二进制大对象(BLOB)存储系统的可靠性和存储效率,设计了一种新的BLOB存取算法.通过建立分布式BLOB存储模型,设计了基于多列等候队列的存储结构,证明了适当增加等候队列数量,可改善存储系统性能.系统采用混合型数据放置策略,分布地存储BLOB数据于FTP群集中各个结点,并将其属性存储于元数据服务器.结果表明,随着系统容量的不断增长,各结点存储性能稳定,数据完整性良好.To promote the reliability and efficiency of storing Binary Large Object （ BLOB ）, a new algorithm of storing BLOB data was proposed. Through the establishment of distributed BLOB storage model, the storage structure based on multiqueue waiting line was presented. It was proved that an appropriate increase in the number of waiting queues can improve system performance. The strategy of hybrid data placement was adopted, BLOB data was stored on the node of distributed FTP cluster, and BLOB data attributes was stored in the metadata server. Experimental results show that, with the constant growth of BLOB storage system capacity, each node has stable storage performance and good data integrity.

关 键 词：BLOB存储 中间件 M/M/1模型 BLOB混合放置策略 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]