(no subject)

Summary of the paper

This paper reviewed ARPANET routing metric, through analysis of SPF
and the revised routing metric. Focused on the delay metric design
especially for heavy traffic conditions, the revised routing made
substantial performance improvements. The main content of the paper
is to present, analyze, and measure the revised metric and its
performance.

At first, this paper summarizes the original algorithm, SPF
algorithm, and delay-SPF. Improved from original one, the SPF
determines the relative appeal of network links based on a cost
associated with each link. The term delay-SPF refers to the case
where routs are computed using SPF and the link metric is measured by
delay. After analyzing the link metric of D-SPF, the paper comes to
the limitations of the delay metric. Under conditions of light
loading, the approach works fine. But on heavily-loaded network,
queueing delay can exert a significant influence on the delay
computed for a link.

Then, paper presents the revised link metric. The modifications
includes three mechanisms that control the change between
successively reported update values for a particular link. The goal
is limited to damping routing oscillations and reducing routing
overhead on link bandwidth and PSN CPU.

In the third part, paper gives the analysis and measurement of the
behavior of SPF, comparing D-SPF and HN-SPF with the conclusion "HN-
SPF is the safety net that compensates for bad network designs and
unexpected changes in traffic patterns. It makes good use of network
bandwidth and can automatically handle variations in traffic that are
several times the the designed traffic level" .

At last, paper gives the conclusion: The HNM has substantially
improved the performance of routing in the in the ARPANET. Under
heavy loads it changes its criteria to give the "average" route a
good path.

Points in favor or against

The paper is generally well written, with fine and clear
presentation. There are two points in the paper in my favor. One is
that author analyzes the problems in the comprehensive view. It
includes the performance both under light traffic conditions and on
heavy-traffic condition. Through this method, author analyzes and
compares the link metric of D-SPF and HN-SPF. The other point is that
for behavior of SPF, author investigates this problem both in
analysis (SPF Model) and in measurement (Equilibrium Calculation).
All these could strongly support the author's viewpoints. Also,
author provides selected results from a study conducted by BBNCC on
the effectiveness of the revised metric in the ARPANET. The table
quotation enhances the conclusion. However, for the point against, in
my view, there are a few grammar mistakes.
Received on Sun Sep 17 2006 - 17:38:36 EDT