On Mon, Oct 26, 2009 at 10:36 AM, Scott Brim <[email protected]> wrote:
> - Loc/ID separation is not directly routing's problem.  The root cause,
>  and the pivot point for a solution, is identification functions that
>  use topology-dependent information as input.  These functions are
>  primarily in endpoints but are also in network infrastructure.  They
>  should be fixed, to the extent they reasonably can.  Where they can't,
>  then routing, mobility, etc. must take up the slack.  The RRG
>  recommends to the IETF that it needs to decide where that line is --
>  what identification functions will they assume will be fixed and which
>  not -- so that other Internet technologies, particularly routing, can
>  have a clearer idea what they have to do.

Hi Scott,

As a group we looked for a definition of locator and identifier and
got a dozen of mutually incompatible answers. I suggest we avoid that
terminology in any recommendation lest it make the recommendation
clear as mud.

Can your statement be rephrased as something similar to:

"layer-4 and higher references to the communicating endpoints are
strongly bound to the packet elements used for layer-3 forwarding
decisions, such that practically speaking one can not change
independent of the other. The pivot point for a solution would be
weakening that binding until it's no stronger than the binding to the
packet's layer-2 elements."

Which is still somewhat opaque because of the circumlocution, but at
least it's built on concepts that are precisely understood.

> - NAT is now architecture.

Concur.

Regards,
Bill Herrin


-- 
William D. Herrin ................ [email protected]  [email protected]
3005 Crane Dr. ...................... Web: <http://bill.herrin.us/>
Falls Church, VA 22042-3004
_______________________________________________
rrg mailing list
[email protected]
http://www.irtf.org/mailman/listinfo/rrg

Reply via email to