[e2e] Can feedback be generated more fast in ECN?

Panos GEVROS P.Gevros at cs.ucl.ac.uk
Thu Feb 15 05:49:35 PST 2001


 |I still think that any solution capable to quickiest notificate congestion
 |must be the matter of our research, even if it imply more cost, for biggest
 |benefits.


it is typical to have more theoretical vs. more industrial positions. true 
enough SQ (with current semantics) is a misnomer for what (me at least) would 
regard (from a research pov) as method for backwards notification - such 
mechanisms have not been sufficiently analysed (probably for pragmatic reasons)

with the danger of repeating myself: the point is that end-points require 
*information* from the network for adaptation. ECN is exactly this : 
information we can afford. Of course a whole packet can carry more information 
and thus may prove more valuable in the long term. (..and it may even arrive 
faster) -this is common sense.
not so much timely delivery as the information content of the feedback signal 
can make backwards notification mechanisms potentially useful,

would it be affordable?  one cannot do any cost-benefit analysis unless there 
is a  concrete proposal, to be weighed against what is regarded as acceptable 
router implementation complexity. i believe there is design space  - unless we 
axiomatically accept that any further computation in the router is prohibitive,

Panos




More information about the end2end-interest mailing list