This should have been gone to the list.

From: Robinson, Dave C (Dave) [mailto:[email protected]] 
Sent: Mittwoch, 05. März 2014 14:41
Subject: RE: Notes

Richard et al,

The notes  fairly represent what I said at the mic.  

However I'd like to reply to Stuart's  response to my commend on ECN and 
fairness.  During the ICCRG discussions Keith Winstein showed some results of 8 
competing flows with RENO where there was significant difference in bandwidth / 
latency between one  flow and the seven others. This is expected. Inside ALU we 
have also shown how two HTTP Adaptive Streaming session on the same congested 
link (e.g. ADSL link into home) can unequally share the available bandwidth 
resulting in different QoE. I'd assert that fairness is a valuable feature of a 
congestion avoidance scheme / AQM. Yes more bandwidth helps. But we find ways 
of filling this bandwidth. I'd also agree that latency is important. But the 
size and variance of latency which can be tolerated depends a lot on the 
application. We need to be clear on the objectives. And these will change on 
whether it is inside the data centre, core of an operators network, in access 
network, or in home network. 

Regards
     Dave

Dr Dave Robinson
IP Routing and Transport
Alcatel-Lucent
+44 7801 878952(m)

    
 
 
 

_______________________________________________
aqm mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/aqm

Reply via email to