Review for The Revised ARPANET Routing Metric

Gennady Pekhimenko.

Summary.

This paper describes the performance improvements of ARPANET that were
made for user delay (new algorithms and metrics) and effective network
capacity. We see how situation is changing from the first used routing
algorithm (Bellman-Ford shortest path) till greatly modified one (Hop
Normalized SPF). The last one overcomes most of the problems for the
previous algorithms :
-- routine oscillations and reduction of effective bandwidth(D-SPF)
-- non-average link metrics and formation of persistent
loops(Bellman-Ford)
and saves desirable features like destination-based addressing and
dynamically routing.

A good result of this work is analyzing the main problems of current
algorithms like routing oscillations and the difference between light
and heavy traffic loading. They understood the main reasons for these
problems and proposed the new revised link metrics with normalization.
Then paper gives us some practical measurements for the SPF model they
use and some explanations for them. But sometimes it's hard to
understand why they get such a result. I think some of the results
needs more explanation and additional theoretical research to be sure
in these results.
Received on Tue Sep 19 2006 - 01:49:04 EDT