From: Kate Everitt (kteveritt@yahoo.com)
Date: Mon Nov 01 2004 - 07:31:00 PST
A. Khanna and J. Zinky, "A Revised ARPANET Routing
Metric," SIGCOMM 1989.
Review: Katherine Everitt
This paper describes the behviour of various routing
schemes under different traffic conditions. It
enumerates difficulties encountered with the SPF
(Shortest Path First) routing scheme, which works
well for light to moderate conditions but does poorly
with heavy traffic. IT is an improvement over
Bellman-Ford SPF because it has knowledge of the
network topology, however it still uses a delay based
metric, which can cause oscillations. During times of
high congestion, instead of the simple shortest path
first which causes problems, they employed a capacity
based metric which takes into account how the network
will respond to the changes. This new scheme, HN SPF
limits the range of values that will be outputted to a
function of the delay, so that extremes will not cause
dramatic changes and network instability.
The main problems with this paper is that the
constants chosen seem a bit arbitrary. (Eg 56kbs
satellite lines are twice as expensive as 56kbs
terrestrial lines.) Sure, they give a 46% reduction in
RT delay even with a 13% increase in throughput.
However, the author don’t give good justification or
heuristics for determining what these values should
be. This means that in a more heterogeneous network,
it’s unclear if working values could be found or what
they should be.
__________________________________
Do you Yahoo!?
Read only the mail you want - Yahoo! Mail SpamGuard.
http://promotions.yahoo.com/new_mail
This archive was generated by hypermail 2.1.6 : Mon Nov 01 2004 - 07:31:06 PST