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

[Tim Moors]

Bob Braden wrote:

>What information should the network be giving
>a reliable E2E transport protocol like TCP, for any conceivable
>network/link layer variants?

This question follows a discussion amongst the Transport Area Working Group
(ftp://ftp.ietf.org/ietf-mail-archive/tsvwg/) over the last couple of days.
The discussion started with a question about why the Tahoe release of TCP
"performed slow start only if the other end was on a different network".
Part of the motivation for the question was a renewed desire for this sort
of behavior in limited circumstances when the end-to-end path does not
traverse routers, e.g. a single hop across a wireless network, or across an
optical circuit.  One thread of the discussion questioned how an instance of
an e2e protocol like TCP can determine whether its communicating peer is
"local".
Should the network provide this sort of information to the e2e protocols?
This leads into Bob's broader question.

What do end2end-interest list readers think?


Tim Moors
___________________________________
Web: http://uluru.poly.edu/~tmoors/