[e2e] Mixed ECN and Non-ECN traffic flows.

[Chong Poh Kit]

On Fri, 11 Oct 2002 09:33:36 -0700, Christian Huitema wrote:
>>I've done some simulations using a mixture of ECN and non-ECN capable
>TCP
>>flows over NS-2 and I've discovered that at a bottleneck link with a
>>single
>>FIFO queue, ECN flows tend to grab more than their fair share of the
>>bandwidth when an AQM (RED,BLUE) is used at the router.
>
>This looks like a good thing to me. After all, we need some incentive
>for people to deploy ECN... As long as the ECN capable flows do not tend
>to actually exclude the non ECN flows, I believe that we are fine.
>
>-- Christian Huitema
>

My view is that the transition process from non-ECN to fully ECN TCP will be a long and gradual one.
I've found that in certain conditions (high loads, low latency) the avg perflow throughput for ECN traffic is almost 2X that of non-ECN 
traffic flows.
Punishing users who are stuck with legacy software would seem to be unfair. I noticed that you worked for Microsoft and since most of the 
installed OS on end hosts are Microsoft Windows, could you tell me if ECN support would need a simple patch or a whole upgrade of the OS 
(if it's not a company secret)? If an OS upgrade is needed, I would doubt that many users in poorer countries, where bandwidth is limited, 
could afford an OS upgrade.
Wouldn't it be important to protect the non-ECN traffic from ECN traffic while the transition process is ongoing?

Regards,

P.K. Chong
Post-graduate Student
Multimedia University, Malaysia
6:16:55 AM 12/10/2002