Re: geo short vs long term? [Re: Geo pros and cons]

[Iljitsch van Beijnum <iljitsch@muada.com>]

On woensdag, apr 16, 2003, at 15:10 Europe/Amsterdam, Sean Doran wrote:

> > I am asking you here and now to step back and let someone else, who 
> > can take all of this seriously, take over.
> I would seriously appreciate it if someone would volunteer himself or 
> herself to the ADs so that I am freer to address (apologies for the 
> pun) many of the discussions here in a much more me-like fashion.
Sounds good to me.

If nobody wants the job, maybe the ADs can find someone with no 
previous multi6 experience? A fresh point of view certainly wouldn't 
hurt.

Now that we have that out of the way...

> > Constrained L4 architecture, anyone? Multiaddressing needs a huge 
> > amount of work before it can catch up with the level of functionality 
> > provided by today's routing-based approaches. That's why we're still 
> > discussing the routing stuff, even though we know it has inherent 
> > scalability problems.
> Do you think the WG could arrive at a consensus on exactly these 
> points?
> That is: multi6 could, if there was agreement,  document that the L4 
> architecture as it stands now is in the way of considering solutions 
> which we do not at present "know ... has inherent scalability 
> problems".
> I happen to favour such a recognition/admission of the problem.
> Whether this would trigger useful reactions outside multi6 is a point 
> which
> interests me enormously.
Obviously I feel it would still be useful to pursue short term 
solutions in order to buy time to develop good long term solutions, but 
that wasn't the question...

It seems to me that if there is going to be consensus about _anything_ 
it has to happen during a meeting, where there is much more pressure to 
do _something_. On the list, calls for consensus usually don't yield 
much result one way or the other.

But to really answer your question: I think it should be doable to find 
consensus on the fact that layer 4 choices in the past are making our 
life difficult today, but I'm not sure we can agree on a plan of 
attack. I suspect that many people will take the position "if TCP wants 
a stable address, we'll implement magic in the IP layer to make this 
happen" rather than "let's improve TCP". Well, maybe that could be our 
consensus. I'd rather see it happen cleanly by changing TCP, but if we 
can get it to work without too big a mess in IP, I suppose I can get 
behind that.

If we can indeed find rough consensus on that, maybe we don't even have 
to throw it back outside multi6.