[Iccrg] Congestion control definition and requirements of a new protocol

[Mark Handley]

On 2/25/06, John Leslie <john at jlc.net> wrote:
> S. Keshav <keshav at uwaterloo.ca> wrote:
> >
> > A. Definition of Congestion:
> >
> > Network congestion is a state of degraded performance from the
> > perspective of a particular user.
>
>    This really won't do.
>
>    We need to define something we can measure.

And this of course is at the core of the problem.  We do need to
define something we can measure to be able to objectively compare two
solutions.  However, unfortunately the ideal solution is one that
maximizes the utility, which is a subjective measure.

Thus I agree it's useful to define something we can measure, but it's
important to always keep in mind that *all* such metrics are only
rough proxies for what we would actually want to measure but can't.

 - Mark