Re: [Cake] [PATCH net-next v12 3/7] sch_cake: Add optional ACK filter

[Sebastian Moeller <moeller0@gmx.de>]

Hi Kevin,


> On May 18, 2018, at 13:18, Kevin Darbyshire-Bryant <kevin@darbyshire-bryant.me.uk> wrote:
> 
> 
> 
>> On 18 May 2018, at 05:27, Cong Wang <xiyou.wangcong@gmail.com> wrote:
>> 
>> On Thu, May 17, 2018 at 4:23 AM, Toke Høiland-Jørgensen <toke@toke.dk> wrote:
>>> Eric Dumazet <eric.dumazet@gmail.com> writes:
>>> 
>>>> On 05/16/2018 01:29 PM, Toke Høiland-Jørgensen wrote:
>>>>> The ACK filter is an optional feature of CAKE which is designed to improve
>>>>> performance on links with very asymmetrical rate limits. On such links
>>>>> (which are unfortunately quite prevalent, especially for DSL and cable
>>>>> subscribers), the downstream throughput can be limited by the number of
>>>>> ACKs capable of being transmitted in the *upstream* direction.
>>>>> 
>>>> 
>>>> ...
>>>> 
>>>>> 
>>>>> Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk>
>>>>> ---
>>>>> net/sched/sch_cake.c |  260 ++++++++++++++++++++++++++++++++++++++++++++++++++
>>>>> 1 file changed, 258 insertions(+), 2 deletions(-)
>>>>> 
>>>>> 
>>>> 
>>>> I have decided to implement ACK compression in TCP stack itself.
>>> 
>>> Awesome! Will look forward to seeing that!
>> 
>> +1
>> 
>> It is really odd to put into a TC qdisc, TCP stack is a much better
>> place.
> 
> Speaking as a user of cake’s ack filtering, although it may be an odd place, it is incredibly useful in my linux based home router middle box that usefully extracts extra usable bandwidth from my asymmetric link.  And whilst ack compression/reduction/filtering call it what you will, will come to the linux TCP stack, as yet other OS stacks are less enlightened and benefit from the router’s tweaking/meddling/interference.

	I believe this is a good point, it is really the asymmetry of the link that makes ACK suppression more or less desirable, and it is quite helpful if the adaptation to that link only needs to be configured on one device. I think this is similar to applying MSS clamping on a router to account for say PPPoE overhead as compared to relaying on path MTU discovery or having to configure the MTU on all end-points.



Best Regards



> 
> 
> _______________________________________________
> Cake mailing list
> Cake@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/cake