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

[Jeremy Harris [RU-UK]]

At 10:34 10/10/01, Tim Moors wrote:
>One thread of the discussion questioned how an instance of an e2e 
>protocol like TCP can determine whether its communicating peer is
>"local".

At least in so far as affects whether congestion-control is needed
(and by extension, TCP slow-start) the particular aspect of "local"
needed is "no opportunity for congestion".  If the link is a dedicated
wire, fine.  If a 10Mb ethernet _and_ the TCP gets backpressure from
the ethernet driver, still fine.  Put anything with a queue in the path
(eg a router, or an [over]intelligent switch), not fine.  Put anything
that fans-in several links to one (10Mb - 100Mb switch) or bridges to
a lower-bitrate link, without signalling strict backpressure, not fine.

The current state is just that we cannot tell when such devices are
_definitely_ not in a path.

- Jeremy