From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-pj1-x102d.google.com (mail-pj1-x102d.google.com [IPv6:2607:f8b0:4864:20::102d]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by lists.bufferbloat.net (Postfix) with ESMTPS id 3470E3B29E for ; Mon, 11 Sep 2023 12:23:40 -0400 (EDT) Received: by mail-pj1-x102d.google.com with SMTP id 98e67ed59e1d1-26f7f71b9a7so3767766a91.0 for ; Mon, 11 Sep 2023 09:23:40 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20221208; t=1694449419; x=1695054219; darn=lists.bufferbloat.net; h=cc:to:subject:message-id:date:from:in-reply-to:references :mime-version:from:to:cc:subject:date:message-id:reply-to; bh=0bfkPqgM9L5AigSpcd6dP/VenRK/q/SdABgCp9Y5HLA=; b=LjGQqNOPIklCTXGcL0XulNgmpfE2sdvkSilmIvTJIENWTGH46u/7TDbswXxUT4G2nr 6tIDhzZvSzYbNy2aqHAotRhLwTPAzl2E7mYjTgb64axpsQ87ePKyBtHigCdAfAZBDB6r oR7BDLi6ekO/SMkS9HjR/nmGmMxzQUTfT/c67/mEt+Lp/VTJ2DCP6Xt6b5+MArFVg5EW SxN7d2tHNx2Dobv5Lot8UrWdbNfYLG7B2VAOVRZg19T1hfvxzZZnKGaHpZ6XSq0jmA1B uyuXjl1x3O4OhjpeqqKkPvPn5/slG2ezNS0dwP6ZkYLdATdok1QIOzYRYbZsA8zwY2Wi rI6g== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1694449419; x=1695054219; h=cc:to:subject:message-id:date:from:in-reply-to:references :mime-version:x-gm-message-state:from:to:cc:subject:date:message-id :reply-to; bh=0bfkPqgM9L5AigSpcd6dP/VenRK/q/SdABgCp9Y5HLA=; b=MmWCxBSeDPeKOOr0qPaUt4HhyA7D+NENYZlGSPDOOWTeoNLarIrL3l5Z9Ofl/FdPk9 U4E6Io+zrjDRRAZv+Fc19uSdWjnVCfz2j+UK09IhjfKp6EeYMAahK7tCcxOJKnosb+di ijyRU1+imSM//tRjG+1tAbedozeDQc3SIVulRmBvYbgm2t+E//suz8BBUZxOpPlWydzT s4ntx/fw5PxktTjQ8hEVorzrREHNmJD5TyQXWYtsiJrCnnQ+i1qXqkLcAFZCCubeOU6l XetmRrhar1yVEcwdmBOq1xRiL7hKWffKBC0aTIqNgY0cLfx5hgzmoM8oJW4uWAKrE3Kg 41aQ== X-Gm-Message-State: AOJu0Yy2LmZ6WiHCWoQcmPUU8Ihf6q0v8UawHUt6yURx9F+H6HCTb2Lg wFtdnW0y2aTvjpUkeN2LiIOcP5saBCzDN3X0Dug= X-Google-Smtp-Source: AGHT+IFhzM3ZWT2+qs/sDE20Xf5UhICP68woiO4G76tEec7l0aHbqdsgBc1vGTYhWXMzVqUY0aTKNlJOlUMq8DPY30w= X-Received: by 2002:a17:90a:9f01:b0:26c:e606:f455 with SMTP id n1-20020a17090a9f0100b0026ce606f455mr9777596pjp.36.1694449418947; Mon, 11 Sep 2023 09:23:38 -0700 (PDT) MIME-Version: 1.0 References: In-Reply-To: From: Dave Taht Date: Mon, 11 Sep 2023 09:23:25 -0700 Message-ID: To: Ignacio Ocampo Cc: libreqos@lists.bufferbloat.net Content-Type: multipart/alternative; boundary="000000000000b4d21d060517bfcb" Subject: Re: [LibreQoS] LibreQoS upstream + Cake/FQ_Codel in CPE Router? X-BeenThere: libreqos@lists.bufferbloat.net X-Mailman-Version: 2.1.20 Precedence: list List-Id: Many ISPs need the kinds of quality shaping cake can do List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 11 Sep 2023 16:23:40 -0000 --000000000000b4d21d060517bfcb Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable On Mon, Sep 11, 2023 at 9:18=E2=80=AFAM Dave Taht wro= te: > Most of our users are in the libreqos chat, and you will in general get > fasters responses there than here. > > > On Sun, Sep 10, 2023 at 7:35=E2=80=AFPM Ignacio Ocampo via LibreQoS < > libreqos@lists.bufferbloat.net> wrote: > >> Hi all, >> >> Is it recommended (or not), to run Cake/FQ_Codel in CPE routers, even if >> LibreQoS is running upstream? >> > > We recommend running cake on the CPE up, and LibreQos to manage the down. > It is very helpful to have cake running in nat mode on the actual device > doing nat (and in that case managing the down also at that device can hel= p) > > We recommend using wifi routers that run fq_codel natively on the wifi, > also, as at higher rates the bottlenecks' bloat shift to the wifi. > > https://lwn.net/Articles/705884/ > > Many devices support fq_codel on the wifi natively, the eero 5, evenroute= , > and many gl.inet products do. > >> >> If so, which are the advantages? >> > > Running cake on the actual bottleneck prevents malignant traffic from > escaping the home network. A mere ping flood can be controlled by libreqo= s, > but it is better to slow that down and mix it up with all the other traff= ic > at the cpe. The algorithms for ack-filtering and congestion management ar= e > always more accurate when put on the actual bottleneck. > To clarify this a bit further, by default libreqos ships with two parameters that actually override the overall settings somewhat. We multiply by 1.09 by default to allow for sufficient slop to look good on speedtest, and have another parameter that helps modify the per hop transit settings. What you were sort of suggesting below is conciously underprovisioning the cpe, where by default we already overprovision libreqos. > > Also, I'm assuming I will have to set up the router with a lower bandwidt= h >> than the assigned by LibreQoS? >> > > yes. > >> >> For instance, if LibreQoS is assigning 13Mbit to the customer, the route= r >> (with Cake or FQ_Codel) enabled need to have a limit of ~11Mbit? >> > > On up from the cpe traffic, you should be able to use the same number as > you use in libreqos. For shaping the down at the cpe the historical > recommendation has always been 85-90%, but with libreqos managing that, n= o > shaping of the down is needed, although I do call out the nat case as a > potential exception. > > >> >> Please advise. Thanks! >> >> -- >> Ignacio Ocampo >> >> _______________________________________________ >> LibreQoS mailing list >> LibreQoS@lists.bufferbloat.net >> https://lists.bufferbloat.net/listinfo/libreqos >> > > > -- > Oct 30: > https://netdevconf.info/0x17/news/the-maestro-and-the-music-bof.html > Dave T=C3=A4ht CSO, LibreQos > --=20 Oct 30: https://netdevconf.info/0x17/news/the-maestro-and-the-music-bof.htm= l Dave T=C3=A4ht CSO, LibreQos --000000000000b4d21d060517bfcb Content-Type: text/html; charset="UTF-8" Content-Transfer-Encoding: quoted-printable


=
On Mon, Sep 11, 2023 at 9:18=E2=80=AF= AM Dave Taht <dave.taht@gmail.com= > wrote:
=
Most of our users are in the libreqos cha= t, and you will in general get fasters responses there than here.

<= /div>

On Sun, Sep 10, 2023 at 7:35=E2=80=AFPM Ignacio Ocampo via LibreQoS &l= t;libre= qos@lists.bufferbloat.net> wrote:
Hi all,

Is it = recommended (or not), to run Cake/FQ_Codel in CPE routers, even if LibreQoS= is running upstream?

We recomm= end running cake on the CPE up, and LibreQos to manage the down. It is very= helpful to have cake running in nat mode on the actual device doing nat (a= nd in that case managing the down also at that device can help)
<= br>
We recommend using wifi routers that run fq_codel natively on= the wifi, also, as at higher rates the bottlenecks' bloat shift to the= wifi.=C2=A0


Many devices support fq_codel on the wifi natively, the ee= ro 5, evenroute, and many gl.inet products do.

If so, whic= h are the advantages?

Running = cake on the actual bottleneck=C2=A0 prevents malignant traffic from escapin= g the home network. A mere ping flood can be controlled by libreqos, but it= is better to slow that down and mix it up with all the other traffic at th= e cpe. The algorithms for ack-filtering and congestion=C2=A0management are = always more accurate when put on the actual bottleneck.=C2=A0

To clarify this a bit further, by defa= ult libreqos ships with two parameters that actually override the overall s= ettings somewhat. We multiply by 1.09 by default to allow for sufficient sl= op to look good on speedtest,
and have another parameter that hel= ps modify the per hop transit settings.

What you w= ere sort of suggesting below is conciously=C2=A0underprovisioning the cpe, = where by default we already overprovision libreqos.=C2=A0

Also, I'm assuming=C2=A0I will have to set up=C2=A0t= he router with a lower bandwidth than=C2=A0the assigned by LibreQoS?
<= /div>

yes.=C2=A0

For insta= nce, if LibreQoS=C2=A0is assigning 13Mbit to the customer, the router (with= Cake or FQ_Codel) enabled need to have a limit of ~11Mbit?

On up from the cpe traffic, you should be able= to use the same number as you use in libreqos. For shaping the down at the= cpe the historical recommendation has always been 85-90%, but with libreqo= s managing that, no shaping of the down is needed, although I do call out t= he nat case as a potential exception.
=C2=A0

Please advise. Thanks!

--
Ignacio Ocampo

_______________________________________________
LibreQoS mailing list
LibreQo= S@lists.bufferbloat.net
https://lists.bufferbloat.net/listinfo/libreqos


--
Dave T=C3=A4ht = CSO, LibreQos


--
Dave T=C3=A4ht = CSO, LibreQos
--000000000000b4d21d060517bfcb--