Re: [Cake] LEDE project - piece_of_cake or layer_cake QOS script

[Sebastian Moeller <moeller0@gmx.de>]

Hi Alex,

> On Oct 29, 2016, at 16:17, Alex Garrood <agarrood25@gmail.com> wrote:
> 
> Hi all, I've just flashed a LEDE image to my home router in order to
> make use of Cake.
> 
> I noticed there are two different Queue setup scripts in luci-app-sqm:
> 
> layer_cake.qos:
> This uses the cake qdisc as a replacement for htb as shaper and
> fq_codel as leaf qdisc. This exercises cake's diffserv profile(s) as
> different "layers" of priority. This script requires that cake is
> selected as qdisc.
> 
> piece_of_cake.qos:
> This just uses the cake qdisc as a replacement for htb as shaper and
> fq_codel as leaf qdisc. It just does not come any simpler than this,
> in other words it truely is a "piece of cake". This script requires
> that cake is selected as qdisc.
> 
> 
> Which would you recommend for a home network type setup and are there
> any particular advantages or disadvantages to either option?

	So do you use DSCP to selectively mark the “priority” of packets? If yes you might want to look at layer-cake. Do you really want the computationally cheapest* way to solve your generalized shaping needs have a look at piece-of-cake. For most users I assume that piece-of-cake should be fine. That said, sqm-scripts could do abetter job at exposing cake’s specific options, but since I have lost my faith in cake’s speedy upstreaming I am kind of hesitant to add options to luci-app-sqm that might not survive the contact with linux kernel, and I absolutely detest the unholy non-systematic mess that cake’s overhead accounting options turned into… (and there are ways in the GUI to pass un-checked strings of options to the under laying qdisc’s already).

Best Regards
	Sebastian


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