[e2e] Open the floodgate

[Cannara]

Bob, I wasn't aware of anyone "randomly attacking".  I'm sure anyone who
studies the archives for this list for the last two years or so will see many
issues raised in a technical manner, but with no result other than a
smothering bureaucratic response.  

The "solid architectural foundation" thing is to mean what -- that
everything's ok and only tweaks are needed?  Was adding flow control
(congestion control) at the transport layer in the '80s done for architectural
soundness, or out of raw fear when Internet near collapses were sensed?  Most
of us know the answer to that.  

So, it's sound to have TCP do 'fairness' management in network traffic when
other data transports on the same network don't work with it?  The story that
"most traffic is TCP" ignores the fact that it might well not always be, nor
even be now on some subnets.  Plus it sweeps under the rug the fact that flow
control in networks is the responsibility of network-layer devices and
protocols (they drop the packets), yet even if we all agreed TCP should mess
there, it's been allowed no real information from the network -- a clear
mistake in new, lossy radio nets.  So not only is fairness coverage
fragmented, it's functionally incomplete.  Is this an acceptable result of any
architectural foundation?  What control theory does satisfaction with this
state derive from?

Alex

Bob Braden wrote:
> 
> Alex,
> 
> The fundamental design of the TCP/IP suite is based upon a solid
> architectural foundation.  You may disagree with particular aspects of
> that foundation (and you are encouraged to do so on this list), but its
> supporters are not arrogant for asking that its detractors acknowledge
> and engage with that foundation rather than randomly attacking the
> resulting protocol design.
> 
> Bob Braden