[e2e] e2e principle..where??....

[Jon Crowcroft]

In message <5.0.0.25.2.20010603190244.024287a8 at central.cis.upenn.edu>, "Jonathan M. Smith" typed:

 >>Here's a question I once posed in a debate with you at SIGCOMM in Cannes that I
 >>have never fully resolved in my own mind. Does the end-to-end argument support
 >>routers as they work today in the Internet? For example, why not source routing
 >>as a more "pure" end-to-end strategy? I'm not trying to be antagonistic - I'm
 >>trying to understand whether the routing control plane (which is network 
 >>embedded in
 >>the Internet) in fact complies with the e2e principle.


so i see the e2e principle as another instance occam's razor - 

when choosing what to put in network or transport layer, you want to
haev the simplest solution - putting source routing into the 
end system requires end systems to build topology
maps - to do so thewy talk to the topology (routers) - 

this doesnt reduce the complexity at all (there's no free lunch), and does 
reduce the responsiveness since you ONLY get to react edge-to-edge to local outages 
instead of having local recovery

having the ability to put route _hints_ into packets would be very fine if 
(as we do e.g. in ipv6) we had a mechanism to do it without burdening everyone with
hint processing...

paul francis' phd thesis is jolly good on this topic...

cheers
jon