Re: SIP Addressing Limitations

[William Allen Simpson <[e-mail]>]

> It amuses me greatly that there's anyone who still seriously
> believes that network topology, and geography, are even remotely
> related when it comes to international connections.
>
Apparently I am in good company.  With Rehkter, Li, and the CIDR luminaries.

Normally, your messages have better reasoning.  It continues to distress
me that people reply to messages without actually reading the plan.

This is the last time I will bother replying to non-specific requests
for wording changes.

First, let's look at the actual plan:

      There are approximately 227 assigned countries and territories.
      This table size is 2 orders of magnitude smaller than currently
      handled by IP4 world-wide.

      An effective aggregation scheme would result from interconnection
      of all of the networks within each country.  This may be likely in
      the long term (by fiat if not for practical reasons), but is not
      required for efficient operation of this plan.  Splitting each
      country into 10 disconnected portions would still be an order of
      magnitude smaller than currently handled by IP4 world-wide.

Are any of these numbers in dispute?


> from here) - it certainly wasn't that way a shortish while ago.

A shortish while ago, there was no internet at all.  Let's try planning
for the future, not the past.

> Does anyone really believe that routes from that area will
> form any kind of aggregation anytime in the forseeable future?

The plan does not require that it be so.  It merely takes advantage of
the fact when it occurs.  Dynamically.

> with them.  But Thailand, Malaysia, and Singapore all have
> internet connections, none are connected to any of the others,

According to public CIA reports, all 3 have communications links with
each other, including radio, television, undersea cable, and satellite.

Oh, you are limiting this to "internet" links?  You expect that the
current situation is permanent?  Fine.  Even if each country stays
separate, the improvement in table size is 2 orders of magnitude better
than we currently have, and *never* gets worse.

> west coast).   Last I heard, Japan as a single country couldn't
> even manage to present a unified routing interface to the
> rest of the world.
>
Again, even if Japan and every other country on the planet split into 10
parts which refuse to talk to each other except through a 3rd party, the
plan provides an order of magnitude better than we currently have, and
*never* gets worse.

And since a 3rd party is likely to be geographically close, the plan
will allow aggregation from the point of view of a more distant observer.


> International links are composed almost entirely out of politics,
> often expressed in the form of tarrifs set by various
> tellecommunication authorities, and proximity is rarely any
> kind of indicationas to what makes sense, financially, or
> politically in other eays.
>
Yes.  Which is why the plan clearly states:

      The basic administrative unit is the country.  Past experience
      shows that political considerations often override practical
      concerns in the administration of networks.  This plan seeks to
      align the practical with the political.

> I think you don't very well understand international link
> costs.  It costs me exactly the same to install a link to
> the east coast of the US as it does to the west coast.
> Making links that seem to be absurd is sometimes the most
> rational thing to do, and can often be cost effective.

No, you are incorrect.  Any link that you actually *build* will cost
considerably more from Australia to the US east coast than to the west
coast.

You are talking about combining several shared links into a long path.
These links are actually switched at various points.

Any provider which can use the same links for intra-continental and
inter-continental traffic will be in an enhanced position.  Any provider
that limits its connectivity will get its financial pants beaten, unless
it can charge a hefty price for "enhanced security".

The plan takes advantage of (and encourages) likely future increases in
inter-connectivity.  Most importantly, such increases do *NOT* make
routing and addressing worse (as is currently the case).

> There are several links from Asian countries into US east
> coast receivers.   Oten, even if perhaps not always, there
> are good economic reasons for those.

There are no trans-oceanic cables from any Asian country to the US east
coast listed in any resource I have available.  Perhaps you could be
more specific?

You may be referring to the fact that current US *internet* traffic
from some Asian countries passes through somewhere on the east coast.
That is merely an effect of US "policy" (single interconnect between
providers), and has nothing to do with topology.  That policy is already
scheduled to be obsolete.

Again, the plan treats the country as the basic administrative unit.
For external countries, it doesn't matter that you are connected to the
east coast or the west coast, only that you are connected.

Now, do you have a better plan?  One that has a guaranteed set of
constraints, and a provable maximal routing table size?  One that
aggregates so that an international amateur radio operator sees a very
small table?  One that scales on an interplanetary basis?

[e-mail]