From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-yb1-xb31.google.com (mail-yb1-xb31.google.com [IPv6:2607:f8b0:4864:20::b31]) (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 71FA13B29E for ; Mon, 24 Oct 2022 11:20:40 -0400 (EDT) Received: by mail-yb1-xb31.google.com with SMTP id r3so11392571yba.5 for ; Mon, 24 Oct 2022 08:20:40 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20210112; h=cc:to:subject:message-id:date:from:in-reply-to:references :mime-version:from:to:cc:subject:date:message-id:reply-to; bh=w7lXq8u09lJDC9vaHofkoj0drzCcpzGf1A5hil9OKko=; b=WW1HQ/QsEi6IIT+T/IoCSRp9ZdHIs15iSXdxvUNt8B83zOok6e90Y9iZ0ksEhGVAam yGZErF93/9awDoTL3K7Wq/JNn/pb4I8K0ma192EgYTZaWrbSkMyCCGRm1ePlGFcRYmzi dqsRXzCAF5vGtKx998bWIaYl/7yAJUYplyMTsEOVsDRQ1InjT1S+SDUnT9fjW7Uwm7k+ RwqWru0Ijm2JgdJSdDMstFyI3ceaqXf3e+/fAIyiAPrOBVNdEE9skwB6o/Bktedl0y3L T6AhjFXPsoyMrqUShe1k0IyIVAh1CsJCPwrdDUzdoG0S8CS3Ra2VV2YB7SIHQ2/l9FVF Uowg== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20210112; 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=w7lXq8u09lJDC9vaHofkoj0drzCcpzGf1A5hil9OKko=; b=hpLyXyKN5GQrIXHhqtrBKZQLwbFmvo57qZ1SH+dNbh1oRhUKDqVSyBsmTC75REvrYv /a3/8xiVVab213P16guLOc/K8T1x3bqHysJ0rB3tSlvmHzBoi/ISN7/Lp9+rNFv55YAR qt1J+WdXZv7WA0c3VSQYjCqYIfw32mDhWj46fjjM51614F9V3W/cEoRgf34D6ta3ABS+ ihP/q8cxWJeGjjmn//1r9LCdIaGE7yrfZs5WS4ii/79xjbiCvv5+FHCMxDJreuZk5jiI 6ZXbhbNpLoPx7IVDls5qBFIg8r0y29+uJ75LN1MnD35PKujubZoW/QZOTy7XowHkybi/ q7pg== X-Gm-Message-State: ACrzQf2bu9l5W5qwt9P2d/FgJDOb9rtf+YKyWFVi3TnmK/nT4+kQ1a18 0FkwNRN12fBldIVWdQ4VWCvLeJB4GDX74iWOWmVZTSAt3uU= X-Google-Smtp-Source: AMsMyM7gXeV3YdsoHLpexuRgIk8examObxbUWwZ65sUVAZ+Ef/B6uuHSyXhLFzQ10hkHEUgwpFGnN8S2deQ197K1K7k= X-Received: by 2002:a5b:ac4:0:b0:6bb:e351:d5e4 with SMTP id a4-20020a5b0ac4000000b006bbe351d5e4mr29812672ybr.78.1666624839646; Mon, 24 Oct 2022 08:20:39 -0700 (PDT) MIME-Version: 1.0 References: In-Reply-To: From: dan Date: Mon, 24 Oct 2022 09:20:30 -0600 Message-ID: To: Dave Taht Cc: =?UTF-8?Q?Robert_Chac=C3=B3n?= , libreqos@lists.bufferbloat.net Content-Type: multipart/alternative; boundary="0000000000008a841105ebc9554b" Subject: Re: [LibreQoS] ookla speedtest results? 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, 24 Oct 2022 15:20:40 -0000 --0000000000008a841105ebc9554b Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable Dave, I think that 'bouncing' is more CPU time in the browser. I don't think it's indicative over what's actually happening. On Mon, Oct 24, 2022 at 9:09 AM Dave Taht via LibreQoS < libreqos@lists.bufferbloat.net> wrote: > On Mon, Oct 24, 2022 at 8:02 AM Robert Chac=C3=B3n via LibreQoS > wrote: > > > > I can run an Ookla test in the middle of the night with LibreQoS off fo= r > a bit. > > Groovy. > > > From what I recall - without Libre we see download bloat of 300ms or so= . > With Libre, it's gone. On waveform we see 0ms added bloat each direction > for most clients. > > The numbers report by speedtest bounce around a lot, and they tend to > pick a lower number as a final result than I'd > like. If there's a way to capture a movie of it... > > > > And I like the "monitor only" mode idea. > > Perhaps we could create the HTB tree, but just not attach the CAKE qdis= c? > > Technically even HTB could reduce latency by itself on overloaded APs , > so it wouldn't be true passive monitoring, but it would allow us to compa= re > the before/after of CAKE while still having qdiscs we can point > cpumap-pping to. > > > > Alternatively, maybe we could use eBPF PPing just for this passive > monitoring period? > > > > > > On Mon, Oct 24, 2022 at 8:02 AM Herbert Wolverson via LibreQoS < > libreqos@lists.bufferbloat.net> wrote: > >> > >> Are you looking for a shaped result? That'll show "whatever bandwidth > >> is left on the sector" without Libre, and around the customer's paid-f= or > >> target with Libre (hopefully!). > >> > >> One thing Preseem does well is that their onboarding process includes > >> "leave it in monitoring-only mode" for a week, gathering data before y= ou > >> pull the switch (although most ISPs I've talked to just pull the > switch...) > >> It's quite enlightening to see a before/after, even with only FQ_CODEL > >> as the shaper. > >> > >> We could probably do something similar with a "monitor only" mode that > >> doesn't enable any queues (is there a TC equivalent to "dummy" on > >> FreeBSD?). > >> > >> On Mon, Oct 24, 2022 at 8:52 AM Dave Taht via LibreQoS < > libreqos@lists.bufferbloat.net> wrote: > >>> > >>> I am curious if anyone here has speedtest results with libreqos on an= d > >>> off, and can send a screen shot? > >>> > >>> -- > >>> This song goes out to all the folk that thought Stadia would work: > >>> > https://www.linkedin.com/posts/dtaht_the-mushroom-song-activity-698136666= 5607352320-FXtz > >>> Dave T=C3=A4ht CEO, TekLibre, LLC > >>> _______________________________________________ > >>> LibreQoS mailing list > >>> LibreQoS@lists.bufferbloat.net > >>> https://lists.bufferbloat.net/listinfo/libreqos > >> > >> _______________________________________________ > >> LibreQoS mailing list > >> LibreQoS@lists.bufferbloat.net > >> https://lists.bufferbloat.net/listinfo/libreqos > > > > > > > > -- > > Robert Chac=C3=B3n > > CEO | JackRabbit Wireless LLC > > _______________________________________________ > > LibreQoS mailing list > > LibreQoS@lists.bufferbloat.net > > https://lists.bufferbloat.net/listinfo/libreqos > > > > -- > This song goes out to all the folk that thought Stadia would work: > > https://www.linkedin.com/posts/dtaht_the-mushroom-song-activity-698136666= 5607352320-FXtz > Dave T=C3=A4ht CEO, TekLibre, LLC > _______________________________________________ > LibreQoS mailing list > LibreQoS@lists.bufferbloat.net > https://lists.bufferbloat.net/listinfo/libreqos > --0000000000008a841105ebc9554b Content-Type: text/html; charset="UTF-8" Content-Transfer-Encoding: quoted-printable
Dave, I think that 'bouncing' is more CPU time in = the browser.=C2=A0 I don't think it's indicative over what's ac= tually happening.


On Mon, Oct 24, 2022 at 9:09 AM Dave Taht v= ia LibreQoS <libreqos@= lists.bufferbloat.net> wrote:
On Mon, Oct 24, 2022 at 8:02 AM Robert Chac=C3=B3n via= LibreQoS
<lib= reqos@lists.bufferbloat.net> wrote:
>
> I can run an Ookla test in the middle of the night with LibreQoS off f= or a bit.

Groovy.

> From what I recall - without Libre we see download bloat of 300ms or s= o. With Libre, it's gone. On waveform we see 0ms added bloat each direc= tion for most clients.

The numbers report by speedtest bounce around a lot, and they tend to
pick a lower number as a final result than I'd
like. If there's a way to capture a movie of it...
>
> And I like the "monitor only" mode idea.
> Perhaps we could create the HTB tree, but just not attach the CAKE qdi= sc?
> Technically even HTB could reduce latency by itself on overloaded APs = , so it wouldn't be true passive monitoring, but it would allow us to c= ompare the before/after of CAKE while still having qdiscs we can point cpum= ap-pping to.
>
> Alternatively, maybe we could use eBPF PPing just for this passive mon= itoring period?
>
>
> On Mon, Oct 24, 2022 at 8:02 AM Herbert Wolverson via LibreQoS <libreqos@l= ists.bufferbloat.net> wrote:
>>
>> Are you looking for a shaped result? That'll show "whatev= er bandwidth
>> is left on the sector" without Libre, and around the customer= 's paid-for
>> target with Libre (hopefully!).
>>
>> One thing Preseem does well is that their onboarding process inclu= des
>> "leave it in monitoring-only mode" for a week, gathering= data before you
>> pull the switch (although most ISPs I've talked to just pull t= he switch...)
>> It's quite enlightening to see a before/after, even with only = FQ_CODEL
>> as the shaper.
>>
>> We could probably do something similar with a "monitor only&q= uot; mode that
>> doesn't enable any queues (is there a TC equivalent to "d= ummy" on
>> FreeBSD?).
>>
>> On Mon, Oct 24, 2022 at 8:52 AM Dave Taht via LibreQoS <libreqos@lists= .bufferbloat.net> wrote:
>>>
>>> I am curious if anyone here has speedtest results with libreqo= s on and
>>> off, and can send a screen shot?
>>>
>>> --
>>> This song goes out to all the folk that thought Stadia would w= ork:
>>> https://www.linkedin.com/posts/dtaht_the-mushroom-song-activity-6981366665= 607352320-FXtz
>>> Dave T=C3=A4ht CEO, TekLibre, LLC
>>> _______________________________________________
>>> LibreQoS mailing list
>>> LibreQoS@lists.bufferbloat.net
>>> https://lists.bufferbloat.net/listinfo/l= ibreqos
>>
>> _______________________________________________
>> LibreQoS mailing list
>> LibreQoS@lists.bufferbloat.net
>> https://lists.bufferbloat.net/listinfo/libre= qos
>
>
>
> --
> Robert Chac=C3=B3n
> CEO | JackRabbit Wireless LLC
> _______________________________________________
> LibreQoS mailing list
> Li= breQoS@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/libreqos<= /a>



--
This song goes out to all the folk that thought Stadia would work:
https://www.= linkedin.com/posts/dtaht_the-mushroom-song-activity-6981366665607352320-FXt= z
Dave T=C3=A4ht CEO, TekLibre, LLC
_______________________________________________
LibreQoS mailing list
LibreQo= S@lists.bufferbloat.net
https://lists.bufferbloat.net/listinfo/libreqos
--0000000000008a841105ebc9554b--