What TCP/IP protocol headers can tell us about the web

摘要: We report the results of a large-scale empirical study web traffic. Our is based on over 500 GB TCP/IP protocol-header traces collected in 1999 and 2000 (approximately one year apart) from high-speed link connecting The University North Carolina at Chapel Hill to its Internet service provider. also use set smaller NLANR repository taken approximately same times for comparison. principal this are: (1) data suitable constructing traffic generating models contemporary traffic, (2) new characterizations TCP connection usage showing effects HTTP protocol improvement, notably persistent connections (e.g., about 50% objects are now transferred connections), (3) content structure that reflect influences "banner ads," server load balancing, distribution. A novel aspect demonstration relatively light-weight methodology passive tracing only headers off-line analysis tools can provide timely, high quality hope will encourage more researchers undertake on-going collection research community with rapidly evolving characteristics