WWW业务访问特性分布研究

WWW业务表现为一系列的访问序列。而Web Server和Proxy Server的日志很好地记录了这种访问序列的过程及特性。WWW业务的特性研究是Web Server、Web中间件研究和人工合成Web负载的基础。分析了一个Web Server和两个Proxy Server的日志，重点研究了Web页面请求的概率分布、Web静态文档大小的概率分布（含传输文档）、Web静态文档的访问距离的概率分布，并将分析结果同相关文献的结果进行了对比，同时通过试验证实了在使用Size作为Web缓存替换依据时，还应该考虑Web文档的访问频率。

WWW TRAFFIC ACCESSS CHARACTERISTIC DISTRIBUTION RESEARCH

WWW traffic can be treated as a series of access sequences. The logs of Web server and proxy server have recorded this process and its characteristics. In this paper, logs of Web server and proxy server are analyzed. Focus is mainly on Web page request distribution, Web static document size distribution (including transferred document), and Web static document access distance distribution. The result of analysis is compared with that of related literature. Because of the cultural difference, some results are different with literatures, such as Web page request distribution of proxy server logs, Web static document access distance distribution. Relationship between some characteristics is also discussed. An experiment is used to validate a fact: when size is used as the replacement key, Web document access frequency should be considered.