摘要
在大规模数据服务应用中,尤其是备份与共享系统中,服务器往往会面临大量的上行数据流,而对这些数据的处理需要进行频繁的写操作,由于服务器数据存储位置的随机性,在多用户多任务环境下磁盘性能大大降低。传统的缓存技术只是针对小范围数据的局部性进行优化,对于并行备份数据流的效果不好,因此提出一种数据缓冲方法,通过数据封装,对上行数据流进行整理,动态调整IO次序,实现异步的数据处理接口,提高了文件服务器的存储性能。
In the large-scale data service applications, particularly backup and file-sharing systems, the server often face a large number of uplink data stream. In a multi-user environment, the processing of data requires frequent write operations. Due to the randomness of data store locations, disk performance reduced greatly. The purpose of the traditional buffeting technology is optimizing storage according to data locality. But it＇ s not efficient for parallel backup data stream. So a data buffering method is presented, which rearrange the data stream and dynamically adjust the IO order to implement asynchronous data processing interface, and improve the storage performance of the file server.
出处
《计算机工程与设计》
CSCD
北大核心
2010年第18期3946-3949,共4页
Computer Engineering and Design
基金
国家自然科学基金项目(60873246)
教育部创新工程重大项目培育基金项目(708075)
国家教育部博士点基金项目(20070610032)
关键词
数据缓冲
上行数据流
远程备份
异步存储
文件服务器
data buffering
uplink data stream
remote backup
asynchronous storage
file server
