RE: PI/metro/geo [Re: The state of IPv6 multihoming development]

["Tony Hain" <alh-ietf@tndh.net>]

Randy Bush wrote:
> </ad hat>
> 
> > Operational issues in the ISP space have always favored restricting 
> > topology or the knowledge about unaggregated parts thereof.
> 
> not at all.  they have merely been focused on maintaining the 
> knowledge of topology by keeping routers from melting down.  
> and this is not theory, we lived it.

We are in violent agreement here... Keeping the routers from melting
down leads to favoring approaches that keep the topology or knowledge
bounded. 

> 
> that most schemes folk have for new topologies have not been 
> realistic in terms of routing table expansion, should not be 
> taken as isps wanting to suppress information.  that you 
> can't holiday on mars this weekend is not my fault.

I am not arguing that the ISP is wrong for wanting to cost optimize, or
simply stay afloat, just that there are reasons behind the demand for
the 'unrealistic topologies' and those reasons are just as valid even if
the proposed topologies are not. 

> 
> information reduction has been is the only tool the isps have 
> to deal with routing bloat and weak vendor hardware.  if you 
> would care to suggest others, i am all ears.

I am not attacking ISPs or their motives. I was asking the simple
question, do we have the representative voices that can tell us the cost
to the edges of that information reduction? Yes it is the only hammer we
have, but we can choose to tap lightly, swing hard, or something in
between to achieve a balance that keeps the routing system intact while
also allowing the edges to get their job done.

Tony


> 
> randy
>