One note:
Ian McDonald wrote:
I too agree strongly with the parts about TFRC being passed upon
TCP-Reno which is not used in modern stacks in it's original form.
When I use iperf on modern TCP variants I see significantly higher
throughput than the TCP throughput equation predicts. This does mean
that TFRC is using less than it's fair share as you say.
TCP-BIC is a nonstandard TCP modification with significantly more aggressive
behavior. Whether or not CCID3 gets less than TCP-BIC gives no information on
fairness. Maybe TFRC gets less than its fair share, maybe TCP-BIC gets MORE
than its fair share, probably both. The fairness gold standard at the moment
is the most aggressive currently IETF-standardized TCP variant, which is
SACK+LT+etc. CCID3 should not get less than SACK+LT under the same conditions.
Eddie
P.S. I personally don't think EXACT fairness is THAT important, and maybe
eventually BIC or a variant will be standardized, but it is important to
understand what this working group should strive for.
