网络存储I/O流水机制研究  

Research on Mechanism of Network Storage I/O Pipelining

作　　者：曾令仿[1] 冯丹[1] 周可[1] 王芳[1] 

机构地区：[1]华中科技大学计算机学院信息存储系统教育部重点实验室,湖北武汉430074

出　　处：《小型微型计算机系统》2006年第1期42-45,共4页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(60303032)资助;国家"九七三"计划项目(2004CB318201)资助

摘　　要：将I/O请求处理划分为多个阶段,为流水线技术引入网络存储提供了新思路.同时,I/O请求处理的各阶段必定通过缓存(内存)来传递或处理数据.I/O请求处理除了直通方式外,大部分依赖存储转发方式(例如对I/O命令的聚散、排队操作等),存储转发的方式下的网络存储I/O流水线具有一些新的特点及其自身特有的制约因素.探讨I/O流水机制,对提高网络存储系统整体性能,具有一定的指导及实践意义.It is a novel idea that an I/O request is disposed in multi-stage by introducing pipelining technology. And some buffers （or RAM） must be used to transfer data from one stage to the other because of having to operate those data （e. g. , gather/scatter technology, queue multiple I/O requests of network on the logic disk）. Store-and-forward is a typical pattern besides cut-through. For store-and-forward, network storage I/O pipelining has some new characteristic and restricted factors. In this paper, mechanism of I/O pipelining is provided which may have significant direction for enhancing performance of network storage system.

关 键 词：网络存储 I/O流水线 I/O瓶颈 

分 类 号：TP333[自动化与计算机技术—计算机系统结构]