On Mon, 12 Oct 2015, Mikael Abrahamsson wrote:
On Fri, 9 Oct 2015, Greg White wrote:
On 10/9/15, 2:04 PM, "Mark Allman" <[email protected]> wrote:
1) *you* shouldn't be using a mechanism that destroys information for
others
2) *you* don't know where your mechanism will have an impact
3) you claim this might be safe *if* AQM is widely deployed
tl;dr summary: myopia is why we can't have nice things
Too true. DOCSIS would have been much cleaner if we didn't have to deal
with the fallout from the myopic TCP designers. :-P
So I agree that most likely, it's beneficial to have fewer ACKs.
What I think people arguing against this practice are these kinds of issues:
http://blog.dan.drown.org/sb6183-dropping-ipv6-traffic/
I don't think there is a solution that we all can agree on, all approaches
have their benefits and drawbacks. I think the above article just shows how
things can go wrong in very subtle ways.
no question that things can go wrong in subtle ways. If network protocol design
was easy, we would have had a couple protocols designed back in the early days
of computing and there wouldn't be the recent rash of new protocol designs :-)
But too much fear that "something may go wrong somewhere" can prevent any
progress.
David Lang
_______________________________________________
aqm mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/aqm
