Intra/Inter - was Inet-II

Fri Oct 11 02:18:51 UTC 1996

alan at mindvision.com (Alan Hannan) wrote:

>> Nobody cares about AUPs in leaf networks.  AUPs in transit backbones
>> are evil.  Or everybody already forgot NSFNET AUP and the tons of
>> related hackery in routing policies all around the world?

> The interesting thing to me below is the assumption that there is
> an inherent difference between a transit backbone and a private
> leaf network.

Hm.  Hm again.  Should i start explaining why AUP-type traffic restrictions
(i.e. choice of different paths depending on who communicates with whom,
or all-out limited connectivity) are different from peer/transit routing
policy restrictions generally found in commercial networks which are
not interested in AUPs but rather in delivering bits efficiently?

Then, any AUP related to government funding is not going to be traffic
path-based; but rather content-based, which makes it unenforceable by
means other than financial coercion.  Those of us who remember how it
worked with NSFNET would rather not have that experience again.
(Yes, and if you think the government wouldn't drag politics in it
you're mistaken -- i remember that getting traffic from USSR to be
allowed to NSFNET wasn't all that trivial; and thanks to Steve G.).

>I believe Manning makes a good point that an AUP is inherent to a
>network.

Sorry, not.  As is Internet is mostly AUP-free, meaning that nobody
cares about content or what providers do with traffic engineering.

>  We have seen an increase in the discussions of AUP with respect to
>  backbones (MCI/SL/UU).  The discussions regarding dumping defaults
>  and forced routing to destinations not advertised all centered
>  around AUP.

Please, do not confuse AUPs and routing policies.  Those are very different
beasts.  AUPs are content-based.  Routing policies are purely traffic
engineering.  Routing policies are not a good tool to implement AUPs, and
in fact attempts to do that lead to very kludgy solutions, and leave
lots of loopholes anyway besides creating enormous headaches to
engineering of commercial carriers who are forced one way or another
to peer with AUP-full networks by the same government.  We've already
got NAP peering requirement mess, thank you.

> While the case is there, it is not that strong.  I think Sagan calls it a
>pseudoscientific argument....

I'm not sure why you want to apply scientific criteria to politics
or drag in the ghost of Carl.  Politic science is an oxymoron,
just like Christian Science.  It isn't any worse to tote an ohmmeter as
a sacred device for measuring purity of devoted than to look for
rigid rules in political world governed by emotions.

> Don't fool yourself.  The I-2 is not the "faster Internet".  It is
> a tool to force those pesky free-thinkers to shut up.

>  Maybe.  More likely it's a tool to give Higher Education
>  institutions a QOS independant from the commercial world (also cheaper).

Cheaper?  Have you _ever_ seen anything done by the government that is
cheaper after you count all hidden costs?  If it really can do things
better than market why not start from more worthy causes like government
supplying the foodstuffs?  Oh, sorry, history is already clear on what
happens then, right?  Few dozen millions dead in Ukrain.  I guess people
allow goverments to play with networks and stuff just because it is not
as important as food.  That still does not change the principle.

>  I don't blindly accept the altruistic guise under which it was presented,
>  but I do think there are sig. other reasons beyond government
>  control.

Aw, how cute.  Sounds like first-grader's taking about Granddad Lenin.
Surely smart grown-up people up will show us the Only And True Way,
they sure know better how to spend our money.

--vadim