(no subject)

This paper is mainly about the workload measurement through examining
content delivery from the point of view of four content delivery
systems: HTTP web traffic, the Akamai content delivery network, and
Kazza and Gnutella peer-to-peer file sharing traffic.

For the explosive growth of Internet, a key challenge for Internet
infrastructure has been delivering increasingly complex data to a
voracious and growing user population. Moreover, some new content
delivery systems has been emerged, like P2P file-sharing system, and
the delivery mechanisms are rapidly changing. The understanding of
content delivery on the Internet now requires a detailed
understanding of how these systems are used in practice. The purpose
of the measurement of workload is to determine the characteristics of
these systems and workload in the Internet.

Authors use passive network monitoring to collect traces of traffic
flowing between the UW and the rest of the Internet. The trace
captures two types of traffic: HTTP traffic which can be further
broken down into WWWm Akamai, Kazza, and Gnutlella transfers, and non-
HTTP TCP traffic, including Kazaa and Gnutella search traffic.

After the data analysis, the conclusion is as follows:
P2P traffic accounts for the majority of HTTP bytes transferred.
P2P documents are three orders of magnitude larger than web objects.
A small number of extremely large objects account for an enormous
fraction of observed P2P traffic.
A small number of clients and servers are responsible for the
majority of the traffic in the P2P systems.
Each P2P client creates a significant bandwidth load in both
directions, with uploads exceeding downloads for Kazaa users.
Overall, the bandwidth demands of P2P systems such as Kazaa will
likely prevent them from scaling further. However, an organizational
P2P proxy cache has the potential to significantly reduce P2P
bandwidth requirements.

This paper is well written and is very readable with file structure.
Moreover, it's very easy to follow the trace step by step. The
analysis with sufficient figures and data tables is very solid. The
research result gives up a clearer picture of Internet workload for
content delivery systems. Authors analyze the workload performance
for P2P file-sharing system which is a new mechanism for content
delivery especially. P2P system occupies the majority of the
bandwidth, although it's distributed structure could allow
scalability. I think the future work should be focused on how to
decrease the workload of P2P systems. For example, to reduce the
request rates that are creating that bandwidth. Actually, nowadays,
BitTorrent system only use the central server to deal with the
requests which could alleviate this problem. Another idea is to add
P2P caching mechanism as mentioned in the paper.
Received on Sat Nov 18 2006 - 15:43:59 EST