From: Erika Rice (erice@cs.washington.edu)
Date: Sun Nov 07 2004 - 21:13:14 PST
"An Analysis of Internet Content Delivery Systems" by Stefan Saroiu,
Krishna Gummandi, Richard Dunn, Steven Gribble, and Henry Levy:
This paper presents a case study of traffic patterns on the UW campus.
Traffic is divided into four categories: HTTP web traffic, Akamai
content delivery network traffic, Kazaa peer-to-peer traffic, and
Gnutella peer-to-peer traffic. The analysis in the paper found that the
peer-to-peer traffic, especially the Kazaa traffic, was the source of a
very large percentage of the inbound and the outbound traffic for the
university. The nature of this traffic was also fundamentally
different; it consisted of a small number large audio and video files
rather than a large number of text and image files.
The authors made the interesting point that caching would likely be of
great service for reducing bandwidth requirements of P2P traffic. It
would be useful because of the large size of the files transferred and
because, like with WWW traffic, there are a relatively small number of
popular files. It would be interesting to see what would happen if such
a cache were to be implemented in practice. Would the decrease in
bandwidth usage offset the potential legal liability? The battle
between the technical superiority of caching the data and the legal
superiority of just blocking the offending ports will certainly prove to
be frustrating and interesting.
This archive was generated by hypermail 2.1.6 : Sun Nov 07 2004 - 21:13:15 PST