Review - Measured Capacity of an Ethernet: Myths and Reality

Review from the paper "Measured Capacity of an Ethernet: Myths and Reality"
by Ivan Hernandez

This paper shows, based on measurements, the great performance that
can achieve the Ethernet technology. The paper gets rid of the myths
about Ethernet, created mainly by obtuse interpretations of
theoretical analysis and bad implementations.

First, the authors give us the context of all the parameters that play
a role in the performance of Ethernet, and also, state the performance
measures that we have to notice in order to measure the performance of a
network. With this information the authors review some of the
theoretical studies about Ethernet and highlight the information
useful in evaluate the technology in real applications. Whenever
necessary, the authors emphasize the, either simplifying or
unrealistic, assumptions of the studies, as well as, the variations to
the protocol, and how these impact in the results.

Secondly, the paper presents the elements of the measurement and its
methodology. Then the paper presents several results and its
associated graphs. They pay particular attention to set up the
experiments as in "real conditions", meaning usual deployments in
actual networks. It is important to note the good performance achieved
in each measurement.

Finally, the authors highlight the important lesson from theory and
how these give us important guidelines to get an successful Ethernet
implementation. Nevertheless, they say, the practical experience tells
us how to obtain good performance. So, it is important to get a good
understanding of the theoretical and practical results. The conclusions
are, nowadays well known, Ethernet is capable of good performance and
CSMA/CD is a successful MAC protocol.

In addition to the main results shown, the paper provide us several
important lessons:

1. In order to really understand the theoretical analysis and learn
from these, we have to put attention to their assumptions and used
models, and how these affect their output.

2. The experimental measures are important, they allow us to get
useful information from almost real scenarios. Furthermore, if we mix
the knowledge provided by theoretical and practical measures, we can
maximize the performance of any given technology.

3. We have to get a good understanding of the fundamental concepts of
any given technology in order to get a successful implementation.
Received on Mon Sep 18 2006 - 18:13:57 EDT