[Irtf-rr] Paper on IP Routing and Traffic Engineering

[Roch Guerin]

Apologies for the "advertising" but I thought this might
be of interest to some on the list based on past discussions
on the topic.

I just wanted to send a pointer to a paper we have
just finished and that shows how one can pretty much
achieve the same "optimal" performance as an MPLS
based traffic engineering solution, using some minor
tweaks to existing IP routing and forwarding.

The paper is entitled "Achieving Near-Optimal Traffic
Engineering Solutions for Current OSPF/IS-IS Networks"
A short version will be presented at INFOCOM'2003 and
both the short and long versions are available at
http://einstein.seas.upenn.edu/mnlab/publications.html.

The paper basically shows how one can take advantage
of the fact that many routing entries map to the same
egress routers to approximate optimal routing arbitrary
closely using shortest path routing and, more importantly,
standard packet forwarding engines that load balance
by distributing traffic "equally" on equal cost shortest
paths.  The paper also shows how, from a more practical
standpoint, this can be achieved by focusing on the traffic
associated with a few 100's routes.

Our goal was to shed additional light on what is feasible
within the context of current routing and forwarding mechanisms,
as this is important when deciding whether or not there is
enough justification for a new technology.

Thanks,

Roch