BCP for multisite multihoming

[Iljitsch van Beijnum]

On 22-jul-2007, at 13:37, Nick Hilliard wrote:

> Can we agree to consign the idea of geographic addressing to the  
> scrap heap
> where it belongs, please?  It's not going to work, ever - and we  
> need to
> stop pretending that it has any future.

Obviously geographic addressing will work just fine: if all else  
fails, just ignore the geographic component and use the addresses the  
same way as any other type. The question is whether the geographic  
component is going to buy us anything. The classic argument is that  
topology isn't aligned with geography so it doesn't. However, there  
are only so many sea cables and the world is a big place. Being able  
to ignore more specific routes from the other side of the globe seems  
like something that could come in handy to me.

> It will work the day that the
> Internet (big "I") is operated as a strict tree structure, which is  
> to say,
> never.

Sure, a tree with a single root would be great. But 7000 trees for  
7000 regions with 1 million inhabitants would still be a huge win.  
Within the region you can do flat routing.