[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [RRG] Long term clean-slate only for the RRG?



In einer eMail vom 04.07.2008 01:00:59 Westeuropäische Normalzeit schreibt bill@herrin.us:
>> -MY- point was that Line 1 need not be there at all. It is an
>> identifier which serves no role in the routing. If you get line 1
>> wrong or leave it off entirely your letter will still get to me.
>
> And this is precisely MY point, too:-)
> Ignore line 1 before the letter hasn't reach the egress post office.
> (maybe we should "invent"  MPLS-2 :-).

No, no, a thousand times no. My point was NOT that you can ignore line
1 until condition X. My point was that you can remove line 1 from the
envelope -completely- and the postal routing system still delivers the
letter correctly.
ok. Nevertheless there are certain interesting points on the delivery path, which are worth to think about.


Drawing the analogy back to it's origin, my point was that in a
clean-slate system, node identity doesn't belong at layer 3. It
belongs above layer 3, either in a layer 3b or in layer 4. ONLY
network location belongs in layer 3. And network location is a
fundamentally ephemeral thing; it changes constantly with your node's
geographical movement and with the the ups and downs of the network's
interconnections.

As for MPLS, who wants to argue that MPLS -is-not- a map-encap protocol? Not me.
You are absolutely right. Properly done there were no need for a label distribution mechanism (LDP). A wrt to routing useful data would already be there, without distribution.

On Thu, Jul 3, 2008 at 6:36 PM, Brian E Carpenter
<brian.e.carpenter@gmail.com> wrote:
> And just to be clear, BGP4 routing is not highly meshed.
> 86.3% of active autonomous systems are purely originators of
> routes (stubs), 13.4% also provide transit, and 0.3% are pure
> transit systems. 42% of autonomous systems originate only one prefix.

I wonder what percentage of streets are cul-de-sacs?

Last month, thunderstorms took out power to about a quarter of
Northern Virginia. With most of the traffic lights out and not enough
police to direct traffic, the commute home was horrid. After averaging
about a mile an hour down one street, I pulled out my laptop and GPS
to try to find some neighborhood roads that would get me there. No
such luck: all the turns that didn't dead-end just ended up right back
on the same street.
Huuh.

On Thu, Jul 3, 2008 at 5:04 PM, Scott Brim <swb@employees.org> wrote:
> [Postal address line 1 is] a port selector.

You may be right. Perhaps our problem with scalable routing is that
we've allowed layer-4 identity to leak into layer 3. Fix the layer-4
problem and a clean layer-3 routing/addressing solution comes into
focus.
 
 

Maybe this is an additional problem.
 
Take care Bill while commuting,
Heiner