Re: [tsvwg] New Internet Draft: Congestion Signaling (CSIG)

Hi Abhiram,

I shortened the CC list to avoid mandatory moderation as I an not subscribed to all of the list, I hope this is OK. Please see {SM} below.

> On Sep 14, 2023, at 02:07, Abhiram Ravi <[email protected]> wrote:
> 
> Thank you for your comments. Responses inline.
> [...]
> 
> > Sebastian: What happens if bottlenecks hand out immediate capacity/rate estimates and all greedy flows start increasing their cwin really fast?
> 
> CSIG supports both min(ABW) and min(ABW/C). Using min(ABW/C), the each flow can increase its rate in proportion to the % available bandwidth, so that in aggregate the flows respect the available bottleneck bandwidth with reduced risk of overshoot. See Section 8.1.2. 
> 
> [...]


[SM] Thanks! I was thinking this was intended to be an absolute solution, but reading 8.1 and your reply makes me realize the core is significant risk reduction. Since perfect is often not achievable I think this is a worthwhile improvement. This still leaves issues with uncoordinated flows hammering a link that can result in transiently increased queuing delay, but since each flow would (hopefully) limit its slow-start up to reaching a cwin appropriate for the total available path capacity that would still be a clear win over the status quo with TCP.
It is however unfortunate that this is not going to be usable over the internet any time soon...

Regards
	Sebastian