[e2e] Overly Overlay; Peer to peer is commonplace

David P. Reed dpreed at reed.com
Wed Jan 2 08:35:23 PST 2002


At 10:38 AM 1/2/2002 -0500, J. Noel Chiappa wrote:
>Nobody is saying it is. But you have to impose a naming abstraction
>hierarchy, or the routing will not scale. You can impose such an abstraction
>hierarchy on *any* graph - it doesn't have to be hierarchically organized.

This embodies precisely the problem I have with most routing 
discussions.  Let me try to get at this from my point of view.  The problem 
phrase is "You can impose such an abstraction on *any* graph".   While 
true, it is irrelevant in a deep way, for two reasons:

   1. a communications system, from its endpoints perspective, is the graph of
      capabilities for end-to-end message delivery.  The internal structure
      of the network (links & nodes) is only relevant when it constrains that.
      Confusing internal structure with external function is not modular.  This
      is like saying that two Pentium IVs are different because they have
      different cache sizes or are made in different processes.

   2. The abstraction being imposed (IP's addressing) is intended to apply to
      communications systems that are not graphs at all.  For example, radio
      networks have topology, but in space, time, and frequency domain they
      are continuous manifolds, not graphs.  (recognizing this manifold 
structure
      has come late to those who keep trying to adapt link-level models to 
radio
      networking, though - Procrustes would be extremely proud of Mobile IP).

>Kleinrock/Kamoun shows that when you do so - in an arbitrary graph - you
>still get reasonably optimal paths (which asymptote to the optimal as the
>network gets large).

"Optimal" is a red flag to me, meaning "check the assumptions".  The 
realism of these particular assumptions is questionable, IMHO.  I'm not 
criticizing Kleinrock's work at all - such work is extremely useful because 
it is usually easier to prove something optimal under approximate 
assumptions than to prove "pretty good" under actual real 
circumstances.  But "optimal" is a word that is often used to trump 
discussion (especially when not qualified with the assumptions).

For example, the techniques used to implement 56K bit modems were thought 
to be impossible, because 32 kb/s was "optimal" for a 4 KHz voice telephone 
channel (modeled as an analog channel).   That the problem was in the 
assumptions was hidden by the tools used to argue "optimality".


>     > So "hierarchically derived" topological addresses are just plain wrong.
>
>Sorry, I completely disagree - although perhaps I'm just misunderstanding
>what you mean by ""hierarchically derived topological addresses".
>
>
>     > More relevant, though again as naive as GUID-based routing, is
>     > geotemporal routing.
>
>Sorry, I don't know what that is - can you elaborate?

I mean approaches that recognize that endpoints are embedded in a spatial 
and temporal physical world, and which calculate routes in a distributed 
manner using such information.  (a prehistoric ancestor of such things is 
Landmark Routing, but also included are various radio techniques such as 
MIT's GRID and low-level routing techniques such as spatial coding, joint 
and multiuser detection - depending on the topology of the available 
connectivity manifold and its embedding in 4-space).


>         Noel

- David
--------------------------------------------
WWW Page: http://www.reed.com/dpr.html





More information about the end2end-interest mailing list