Re: My impressions of the Sunday meetings....

[Kurt Erik Lindqvist <kurtis@kurtis.pp.se>]

> > BUT, I think that we need more data, experience and time.
> Oh, for land sakes, how much more "data, experience and time" do we 
> need
> with the current architecture to know that it's a sad mismatch with 
> what
> people are trying to do out in the world? (Ref Iljitsch's wonderful 
> message
> about how "most services run on several hosts (load balancers)", etc, 
> etc.)
>
> I think we've now got plenty of data to say that minor tweaks to the
> existing architecture just ain't gonna cut it. But of course now we 
> get into
> the whole problem I alluded to earlier, which is how to you get a 
> behemoth
> like the IETF to accept something radically new.
I agree that minor tweaks won't do it. But do I know what type of 
problem you are trying to solve? How will larger allocations given to 
ISPs affect the growth rate of prefixes in the DFZ for non-multihomed 
customers? Are we building a solution for N enterprises? How large do 
we think that the allocations to the enterprises will be? Is this 
mostly going to be home DSL users? Will they all have /48s? How fast 
will this pick-up? How will the pricing imposed by operators for 
multihoming solutions affect the pickup rate?

There is a lot more work that needs to be done than just add multiple 
addresses....especially in the enterprise networks. Currently we can't 
even say at what rate multihomed prefixes are coming online with IPv4.

- kurtis -