On Sat, 10 Oct 2015, Joe Touch wrote:
On 10/10/2015 8:45 PM, David Lang wrote:
But I actually view this as a tragedy of the commons situation. Sending
the extra packets doesn't hurt you, but it does mean that you are using
more airtime than you need, and that hurts everyone (including,
eventually, you)
So, basically, you appreciate the tragedy of the commons when it's your
tragedy and their commons, but not the other way around?
(using 'sed')
s/sending the extra/altering the transport/
s/using more airtime/altering E2E semantics/
I actually disagree that there is a significnt difference in the actual E2E
semantics between sending a single stretched ACK and sending the full number of
ACKs back-to-back. The "transmit burst" issue will happen in either case.
But you will notice that in both cases, I am in favor of reducing the number of
packets on the wire. Packets not send can't interfere with other traffic.
If ACK packets are delayed to the point that additional ACK packets have caught
up with them, they are not providing the rate signalling signal that you are
looking for.
Question: Does anyone know if current server TCP stacks implement the TCP byte
counting that RFC3449 talks about?
David Lang
_______________________________________________
aqm mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/aqm
