[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [RRG] Long term clean-slate only for the RRG?



In einer eMail vom 03.07.2008 23:25:53 Westeuropäische Normalzeit schreibt jmh@joelhalpern.com:
The analogy between street addresses and packet addresses for routing
has MANY weaknesses.
Street routing relies on the fact that connectivity is highly meshed.
And that streets are generally free for anyone to use.
Further, street routing relies on the fact that there is a human being
who can learn and adapt his behavior, if the basic routing fails.  If
you can not get to your destination from the obvious exit from the
highway, you can get back on, and try another exit.  or even call for
help as to how to get to the destination.  Packets are not so
intelligent.   (Tunnels and lookup tables approximate many of these same
techniques.  But they are not the same as having a human at the wheel of
a car.)  The many interesting reports of people driving into trouble
because the nav system said to do something foolish are indicative of
what happens if you try to use a street oriented routing system without
an intelligence in place.
This is why we should do research: to do packet !!! routing better than is the current practice as you describe.
 


Yes, once could imagine a world in which connectivity is mandated and
maintained such that street-like routing will work.  But unless we want
to change the regulatory and business structures of base IP
connectivity, we do not live in that world.
The postal address with street number, street, zip,... is only an analogy. I do not propagate postal addresses
for packet routing.
 

(If we are willing to make that degree of change, something like compact
routing becomes far more interesting, as that requires explicit and
predictable resource sharing.)
 
I have read those papers about compact routing. There I learnt about "stretch":-(
I think I can provide much better routing algorithms and concepts (btw, also to improve ipfrr significantly).
But if you want, then my "topology aggregation" is also somehow "compact routing" :-)
 
Heiner