"Selfish Routing"

Fri Feb 14 23:15:13 UTC 2003

Thus spake "Pete Kruckenberg" <pete at kruckenberg.com>
> http://www.scienceblog.com/community/article1018.html
> ---
> This might be easier to understand if it was more technical,
> but I'm only aware of a lot of disabled features on my
> routers that are supposed to in theory do some of these
> things.

And they're disabled because they often result in routing loops, usually
transient but sometimes permanent.  With very careful planning, you can
create scenarios where these features help; however, it's usually cheaper to
add capacity than to improve efficiency when you include engineering and
operational costs.

> Abstractions and analogies aside, is this really a problem,
> and is it really worth solving? Sounds like a lot of
> additional complexity for the supposed benefits.

Some carriers are solving this problem with MPLS-TE, but not the way the
author suggests.

Other than the MPLS-TE solution, I'm not aware of any ISPs that use
congestion- or RTT-based routing.  [E]IGRP is the only IGP with a mechanism
to implement this on a packet level, and experience shows it is unstable in
most topologies.

S