I'd like to second Mike's note, particularly the call for participation.
Learning how to characterize IP clouds with carefully defined metrics of
delay, flow capacity, and packet loss will be real work.
Note, for example, the discussion this weekend of how simple pings
to a loaded router don't yield a good measure of packet-forwarding loss.
The same could be said for trying to measure round-trip delay to a loaded
router.
If you'd like to join the IPPM email list, send mail to
<ippm-request@advanced.org>. Notes expressing an interest in working on
specific kinds of issues would be particularly appreciated. We'll also be
meeting at the upcoming Dallas IETF.
-- Guy