Re: [RRG] Geographic aggregation-based routing is at odds with reality

[Robin Whittle <rw@firstpr.com.au>]

I am replying to Brian and Ricardo.

Hi Brian,

Let me rephrase this:

>> Geographical aggregation is the sort of thing which looks good on
>> paper, but will never be acceptable in the real world.
> 
> In any case, it isn't forbidden, and never has been. There's
> absolutely nothing to stop the ISPs in Klingonville grouping
> together to set up a local IXP, which gets a short prefix from
> the RIR, and then uses it as a metro area prefix for all
> local customers, with whatever policies and configurations
> are needed in the IXP-connected routers.

The above sentence, when properly interpreted in the context of the
rest of my message, means:

   Ideas to achieve a scalable routing and addressing system by
   re-engineering the routing system to work scalably with addresses
   which are assigned according to the restrictions of "geographical
   aggregation" is the sort of thing which looks good on paper, but
   will never be acceptable in the real world.

This is my opinion, of course.  I am stating it as if it is a fact
because I am sure it is true.  I am frustrated at the way
geo-aggregation keeps emerging in this discussion, which I believe
should concern proposals which have the potential to be compatible
with business, security and policy reality.


Hi Ricardo,

How could a scalable routing system which depends upon geographical
aggregation achieve its goals if there are also connections and
address usage which does not respect whatever geographical rules are
set to support this architecture?

I imagine some new architecture could "improve" scalability, such as
perhaps:

  http://www.cs.ucla.edu/~rveloso/papers/giro.pdf

However, if you only have 50% of the Net using the rules, then the
scalability improvements won't apply to 50% of the packets, 50% of
the addresses, routes or whatever.

I am not sure that it would ever be adopted by 50% of providers and
end-user networks anyway.  What would be the benefits for early
adoptors of a new architecture which restricts their addressing and
routing systems in this way?

But even if the new routing technologies and the new geographic
restrictions on what IP address a network could have were adopted by
50% of networks, I don't see how this would solve the routing
scaling problem.

Cutting the problem in half wouldn't be good enough.  We need some
factor of 10 or more (ideally hundreds, thousands or hundreds of
thousands) scaling improvement to cope with the desired growth in
end-user networks with their own portable, multihomable, address
space.  This growth is desired and required by the end-user networks
themselves.

How do you counter my specific critiques that geo-aggregation's
requirements on addressing and paths taken by packets are at odds
with business, policy and security requirements?


 - Robin


--
to unsubscribe send a message to rrg-request@psg.com with the
word 'unsubscribe' in a single line as the message text body.
archive: <http://psg.com/lists/rrg/> & ftp://psg.com/pub/lists/rrg