[e2e] where end-to-end ends

[David P. Reed]

At 07:03 PM 5/22/01 +0100, Jon Crowcroft wrote:


>the ipnrg draft is a product of the First Foundation
>
>Members of the Second Foundation will tell you (if you can find them -
>hint, check on Trantor) that the true ends are the same old ends - the ends
>here are the mission controller programming the robot task, the robot
>actions, and the responses are the feedback from the output of the
>robot taks - just coz there is some bufferening, header compression,
>and hop-by-hop retransmission doesnt stop this being end-to-end - it
>just stops it from being realtime control - you do not need to go off
>planet to go too far for realtime control...try playing jazz or
>classical music over a 2000km network path and then improvise a
>rallentando.

Jon is right - The real question is not whether we send programs to 
Jupiter, but whether it makes good sense to include active computational 
participants in the relay stations on the path between here and 
there.  What good does it do to robot control algorithms running in the 
vicinity of the asteroid belt, rather than at the point of action?  Are 
computers cheaper and more efficient when outside planetary gravitational 
fields?  That would call for active networking ideas.  But I'm afraid that 
gross delay, while it might call for an alternative to TCP, does not 
invalidate the "end-to-end argument".


- David
--------------------------------------------
WWW Page: http://www.reed.com/dpr.html