> On 12 Apr, 2018, at 9:46 am, Mikael Abrahamsson <swm...@swm.pp.se> wrote:
> I sent this to tsvwg where we're discussing the LE codepoint. Since I am now 
> talking queue settings, I thought it might be interesting to get feedback 
> from this group as well on what advice we should give operators.
> Please take into account that I am aiming for what is possible on currently 
> deployed platforms, seen in the field. Not what might be possible on future 
> hardware/software. So available are generally (W)RED per queue and a few 
> queues per customer.
> I am also going to test a 3 queue setup, where each of these groups of DSCP 
> values would go into different queues where LE would perhaps be assured 5% of 
> BW and the rest split evenly between a BE and "everything else" queue. If I 
> did that, I would probably not start dropping LE traffic until 10-20ms buffer 
> fill.

Given the stated constraints, this looks reasonable to me.  Better to have some 
basic AQM than a purely dumb FIFO.

Better still, of course, to have something more modern...

 - Jonathan Morton

aqm mailing list

Reply via email to