| 多Markov链预取模型在网络内存系统中的应用 |
| |
| 引用本文： | 袁清波,孙国忠,陈明宇.多Markov链预取模型在网络内存系统中的应用[J].计算机工程,2007,33(23):105-107. |
| |
| 作者姓名： | 袁清波 孙国忠 陈明宇 |
| |
| 作者单位： | 中国科学院计算技术研究所计算机系统结构重点实验室,北京,100080 |
| |
| 摘 要： | 主机通过高速网络访问远程内存的性能已经达到或远高于访问本地磁盘的性能，通过各种优化手段，网络内存系统的性能能得到更好的提升。该文基于一个Linux网络内存系统(LNMS)，在客户端一级提出了一种新的预取算法m-ppm，该算法发展了多Markov链预取模型，使之更适合LNMS。在LNMS上实现了另2种常用的预取算法以作比较，实验数据表明，m-ppm算法对多用户模式更有效。
|
| 关 键 词： | 网络内存系统 多Markov链预取模型 性能优化 |
| 文章编号： | 1000-3428（2007）23-0105-03 |
| 收稿时间： | 2007-01-15 |
| 修稿时间： | 2007年1月15日 |
Applicedion of Multi-Markov Chains Prefetching Model in Network Memory System |
| |
| YUAN Qing-bo,SUN Guo-zhong,CHEN Ming-yu.Applicedion of Multi-Markov Chains Prefetching Model in Network Memory System[J].Computer Engineering,2007,33(23):105-107. |
| |
| Authors: | YUAN Qing-bo SUN Guo-zhong CHEN Ming-yu |
| |
| Affiliation: | ??Key Lab of Computer System and Architecture, Institute of Computing Technology, Chinese Academy of Sciences, Beijing 100080?? |
| |
| Abstract: | Good performance can be archived when fetching data from remote memory through high-speed network than that from local disk, and performance of network memory system can be improved by kinds of optimization. Based on a Linux network memory system(LNMS), a new prefetching algorithm——m-ppm is proposed to improve the performance in client part. The algorithm extends multi-Markov chains prefetching model to make it more suitable for LNMS. Other two common prefetching algorithms are implemented in LNMS for comparison. Experimental results show that m-ppm method is more effective, especially in multi-user mode. |
| |
| Keywords: | network memory system muhi-Markov chains prefetching model performance optimization |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
| 点击此处可从《计算机工程》浏览原始摘要信息 |
|
点击此处可从《计算机工程》下载全文 |
