[e2e] What should e2e protocols know about lower layers?

Dan McDonald danmcd at east.sun.com
Wed Oct 10 10:58:28 PDT 2001


<SNIP!>
> The LAN is an important component of the current architecture, and 
> differentiates between local and nonlocal for the use of broadcast to 
> avoid centralization of MAC address registration, and the alternative of 
> explicit dispersed registration.

So far, I agree with you, and have no questions.  We need to be careful about
what properties we assume when we distinguish between local and non-local.
Whether a long link on GB ethernet is misconfiguration or not, it's something
to be taken into account.

> While I appreciate the use of congestion control throughout, I still 
> maintain that there are times when the network layer in particular needs 
> to distinguish between local and nonlocal. I see no reason why the 
> transport layer shouldn't be afforded the same opportunity.

Here is my initial (possibly naive) question:

	What _other_ decisions besides congestion control can I make as a
	transport given knowledge of local vs. non-local?

I don't have much to say on this, and look forward to being schooled on this
subject.

And the other (not-so-naive) question:

	What distinguishes local from non-local?

The non-controversial answers (e.g. broadcast/link-enabled multicast) to that
question are few.  The controversial ones should appear (and hopefully cause
spirited debate) in short order on this list.  People who design and develop
link technologies start with answers to this question, and when they add
value, they do so while maintaining at least the appearance of "local" even
if it is anything but.

Perhaps if we can determine more crisply what separates local and non-local,
we can prevent what should be link technologies from thinking they can
replace IP without rediscovering every problem that this community has
encountered over the past 20+ years.  (Naaahhhh, probably not, but one can
hope, yes?)

Dan



More information about the end2end-interest mailing list