g'day,
"J. Noel Chiappa" wrote:
> From: james woodyatt <jhw(_at_)wetware(_dot_)com>
...
> They are *not* doing it because they "want identifiers for their
> machines that are independent of their location in the connectivity
> topology." They are doing it because they want to share their Internet
> access with multiple computers.
Your take is not matched by what other people are saying, which is what I was
going on (but see below).
...
I'd love to hear of some real data to sort this all out.
I'm with you, although there does seem to be a school of thought here
that says that because its violates the end-to-end principle, it doesn't
matter how many people are doing it. In other words "it's wrong, even if
the duck enjoys it..."
Which leads me to my latest heretical thought - given the amount of
energy being expended to unsuccessfully figure out how to make the
end-to-end axiom coexist with today's Internet, how bad would it be to
spend some time reexamining that particular axiom? Suppose rather than
listing the apps that break when addresses are NATed, and using this
list as a reason to not change things, we use that list as a starting
point to examine ways to fix them? Everyone seems to think the problem
is not intractible, but we can't get there from here because we seem to
have a built-in conflict with our starting assumptions. So, for example,
rather than take "end-to-end" to mean "host-to-host", what happens if we
postulate it to mean "first router to last router"? What if we
postulated a hierarchy of identifiers across a network?
As you pointed out earlier, folks talked about separating out the end
point identifier in the past, maybe we should reexamine this suggesion
in light of today's impasse? Maybe what we really need here is to change
one of the axioms and see if it takes us to a non-Euclidian geometric
space?
Nah, let's just continue trying to square the circle....
This whole debate is starting to remind me of the Quebec comedian who
parodied that province's separatists as wanting "an independent Quebec
within a strong and united Canada". Personally, I wont rest until I have
ephemeral, easily changed end point identifiers that are stable and
permanent!
- peterd
--
---------------------------------------------------------------------
Peter Deutsch pdeutsch(_at_)gydig(_dot_)com
Gydig Software
"It is inhumane, in my opinion, to force people who have a
genuine medical need for coffee to wait in line behind people
who apparently view it as some kind of recreational activity."
-- Dave Barry
---------------------------------------------------------------------