From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mout.gmx.net (mout.gmx.net [212.227.17.20]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by lists.bufferbloat.net (Postfix) with ESMTPS id B49093CB37 for ; Fri, 7 Jun 2024 01:36:57 -0400 (EDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmx.de; s=s31663417; t=1717738602; x=1718343402; i=moeller0@gmx.de; bh=L/7KZLPOq0N30N/tXshOm0niXBkB/iokOnSsnGCI2ys=; h=X-UI-Sender-Class:Date:From:To:CC:Subject:In-Reply-To:References: Message-ID:MIME-Version:Content-Type:Content-Transfer-Encoding:cc: content-transfer-encoding:content-type:date:from:message-id: mime-version:reply-to:subject:to; b=dHLXXoJN97q3xa+leuRHbDz5EDciHp71pR2NVPZIaKwpjo4AVhTkFDqtax7pAf3n uF0ymj2UvjICznOI5XpxQ5EqCTwM1oEnoJnsifBDbtbpBd6z/5H8yWxNIYpUzUjLE j3HKgokXN8BbN4wuf0N0Na5QO2ACrGOffyK5knZ151IcwOy/cWS9fUd3lM2RE08rr mSKKlsGi9ISYkxETeeYA+t8vqC6CiI0DFLl3vj68auZEwmxFUhbtAaM/bWN/k7ASO SLMLuXnP7U0t6DFO9idH0dXOh4VTN16zHjjsiQv8jLL9MB1slnktWc5VNdcPiMdem 4lNsHZZogMD22q+XQA== X-UI-Sender-Class: 724b4f7f-cbec-4199-ad4e-598c01a50d3a Received: from [127.0.0.1] ([78.50.66.52]) by mail.gmx.net (mrgmx105 [212.227.17.168]) with ESMTPSA (Nemesis) id 1MGQj7-1sEdSZ15bx-0011CY; Fri, 07 Jun 2024 07:36:42 +0200 Date: Fri, 07 Jun 2024 07:36:36 +0200 From: Sebastian Moeller To: Dave Taht , Dave Taht via Starlink , Stuart Cheshire CC: Starlink , Colin_Higbie User-Agent: K-9 Mail for Android In-Reply-To: References: <438B1BC4-D465-497A-B6BA-700E1D411036@ieee.org> <79C02ABB-B2A6-4B4D-98F4-6540D3F96EBB@ieee.org> <7E918B58-382A-4793-A144-13A7075CA56C@connectivitycap.com> <13rq2389-9012-p95n-s494-q3pp070s497n@ynat.uz> <6qop2p3o-351p-788q-q1q2-86sosnq3rn21@ynat.uz> <3FF32F52-4A93-496B-85FF-00020FA4A48B@gmx.de> <08F6942E-CC08-4956-B92E-CBEC091D86E4@ieee.org> <2F510BD5-2D7E-4A6A-A3DE-C529D14F6FBC@apple.com> Message-ID: <0D6DC1A5-4934-43A3-9F0A-A637E8FD1A76@gmx.de> MIME-Version: 1.0 Content-Type: multipart/alternative; boundary=----5QIWZKRU9V22EXFH1X5H15JGN04ZHV Content-Transfer-Encoding: 7bit X-Provags-ID: V03:K1:Ejz3aHfOwJ+WAxw2k3vRBk2cMGzeLTnrrJ3GbgSN7sovQMCeGXt mKD6mouMWCzvTDEWfPIwAVBIEnA8i1lVgvdsZa7X2OHP3D9IhHvf0Dv2PKDNYmxyIhyF8op 7wzBkE4U4dffIrv7h/cMsD7uEWvSe2kZCBH3NncuDu9JTaTWgHDnvRymXBSSeWvETrrGMGs 9LaxEKkGFtH2gOg3bkWRA== X-Spam-Flag: NO UI-OutboundReport: notjunk:1;M01:P0:VRliH/H1Bm8=;bCWybk2MgXfjcTtIQF529YhZYap 7fV8lplVLByLdV29gUwvuFpITF3JSoXzC42adqnqCyfV7C0TrKzNrn5suf34HQmsHUSB4Qc+4 c6C9mI1KzvSrmdLpJwWv4IpkEKjEeXvbIymw9psV+2wHwzKXQOzXwh6dCrT/uKi69IZglAxYl xZpNdITLICPXUCbGFUnJsG08IDqvMAEpZzZn75LpQwJnGBMcIhptD9hff14RkiPoONnYzwu42 BJlSeMzl54q4Z9IWW4hQ5OTKhMZaz5+yK9+1AOHFTQ/e+OBimFf3rDcxgSFPtHY9u/cKbnK9X BpWRtCCts6AgzJFoZTsDJ35GEFuqzAdW4SvPCrde5vvSkPBu53nFbS1NnTUXPSmUeElCZcgrc pjHSomO4vy87DR9eJ8Y8P0al1qdp9Q8sHKplIG1ZOda7hdAlqi6DnVcPf44DkrXXxNcoTKEwX FLABiPOlVYAZW2BScQLGD+nPNtvYX16kObjqJsICvRjVai25SbMoBLQqJkOyoivoT5QEqKaXV MA49ybCoHYVPW51Y6F/VkOzTe9V6FskQoLEUXeuPH/GV4+BTlRRaaStoluIj8Fnp1S/LENvZv mk5TGq0ujlEygEs9E+G7Cl/bRJ4iWblupWJ+PavXNRBj6SJ7TR+lJrNmUbaxQ71zmRpIgmLEh 3OB1ZLofxsi6bIsYl4eD/3lWcSzreNUO08MFB5lL+3GRDLvtBuKEHilLDhE3M2WLp1fR0Iutr DkqoqQVIgBxg47jjjDS+wS1FdEptRkAcaDbmzN7NJ6LBCC3IPftdOOgelYApF/Lv73ab9uE/d /Lvl8G+xPlOCfLc29r2m0ctZcogjbX0WijVPv5Njab2dU= Subject: Re: [Starlink] The "reasons" that bufferbloat isn't a problem X-BeenThere: starlink@lists.bufferbloat.net X-Mailman-Version: 2.1.20 Precedence: list List-Id: "Starlink has bufferbloat. Bad." List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 07 Jun 2024 05:36:58 -0000 ------5QIWZKRU9V22EXFH1X5H15JGN04ZHV Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable This is pretty impressive=2E=2E=2E and also is a decent counter against the= common argument that at BNG/backbone information rates flow queuing would = be completely infeasible=2E=2E=2E or it might show that big iron silicon is= just inferior to general purpose CPUs On 7 June 2024 04:28:18 CEST, Dave Taht via Starlink wrote: >I occasionally am happy to point out the 150+ isps now running libreqos a= nd >cake=2E=2E=2E the several hundred running preseem and paraqum and bequant= =2E=2E=2E > >As a rule of thumb about 10k wisp subscribers eat around 25gbit=2E This w= e >(libreqos anyway) can do easily on a 1500 dollar whitebox (and we have >pushed it past 60gbit in the v1=2E5 release entering beta shortly)=2E Thi= s is >usually way more capability than any given isp network segment needs=2E= =2E=2E > >The wisps have got fq codel available native in much of their gear too, a= nd >of course starlink on their wifi=2E=2E=2E > >There are probably 60k isps left to go though=2E There are isps still on >docsis 3=2E0=2E I tend to regard these issues nowadays as being demand s= ide as >these solutions are so widely available now=2E=2E=2E > >But with billions being spent to just upgrade to fiber=2E=2E=2E a dark cl= oud >ahead is above 50mbit most of the bloat moves to the wifi=2E=2E=2E and de= spite >eero, openwrt, Google fiber etc that have been getting it right=2E=2E=2E = sigh=2E > >A bright light at the moment there is all the wifi products coming out wi= th >a mt79 chip=2E > >On Thu, Jun 6, 2024, 10:51=E2=80=AFAM Stuart Cheshire wrote: > >> On Jun 4, 2024, at 16:03, Rich Brown wrot= e: >> >> > Yeah=2E=2E=2E I didn't write that as carefully as I could have=2E I w= as >> switching between "user voice" (who'll say 'speed') and "expert" voice = (I >> know the difference)=2E Check it now: >> https://randomneuronsfiring=2Ecom/all-the-reasons-that-bufferbloat-isnt= -a-problem/ >> >> Thanks for doing that=2E >> >> How about also changing =E2=80=9Cnew faster ISP plan=E2=80=9D to =E2=80= =9Cnew bigger ISP plan=E2=80=9D? I >> know that may sound like a slightly weird phrase, but getting people=E2= =80=99s >> attention by surprising them a little can be beneficial=2E If it looks = weird >> to them and that makes them pause and think, then that=E2=80=99s good= =2E >> >> If the hypothetical ISP imagined here were actually willing to offer a >> plan that truly provided consistently *faster* connectivity instead of = just >> more of the same, we=E2=80=99d be very happy=2E The truth today is that= most IPs >> offer *bigger*, not *better*=2E They are selling quantity, not quality= =2E >> >> (I am intentionally not lumping *all* ISPs into the same bucket here=2E >> Some, like Comcast, are actually making big efforts to improve quality = as >> well as quantity=2E Comcast dramatically reduced the working latency of= my >> cable modem during the work-from-home pandemic, and they continue to wo= rk >> on improving that even more=2E I want to be sure to give credit where i= t is >> deserved=2E) >> >> Stuart Cheshire >> >> --=20 Sent from my Android device with K-9 Mail=2E Please excuse my brevity=2E ------5QIWZKRU9V22EXFH1X5H15JGN04ZHV Content-Type: text/html; charset=utf-8 Content-Transfer-Encoding: quoted-printable
This is pretty impressive=2E=2E= =2E and also is a decent counter against the common argument that at BNG/ba= ckbone information rates flow queuing would be completely infeasible=2E=2E= =2E or it might show that big iron silicon is just inferior to general purp= ose CPUs


On 7 Jun= e 2024 04:28:18 CEST, Dave Taht via Starlink <starlink@lists=2Ebufferblo= at=2Enet> wrote:
I occasionally am happy to point out the 150+ isps now r= unning libreqos and cake=2E=2E=2E the several hundred running preseem and p= araqum and bequant=2E=2E=2E

As= a rule of thumb about 10k wisp subscribers eat around 25gbit=2E This we (l= ibreqos anyway) can do easily on a 1500 dollar whitebox (and we have pushed= it past 60gbit in the v1=2E5 release entering beta shortly)=2E This is usu= ally way more capability than any given isp network segment needs=2E=2E=2E&= nbsp;

The wisps have got= fq codel available native in much of their gear too, and of course starlin= k on their wifi=2E=2E=2E 

There are probably 60k isps left to go though=2E There are isps stil= l on docsis 3=2E0=2E  I tend to regard these issues nowadays as being = demand side as these solutions are so widely available now=2E=2E=2E 

But with billions being s= pent to just upgrade to fiber=2E=2E=2E a dark cloud ahead is above 50mbit m= ost of the bloat moves to the wifi=2E=2E=2E and despite eero, openwrt, Goog= le fiber etc that have been getting it right=2E=2E=2E sigh=2E

A bright light at the moment there is= all the wifi products coming out with a mt79 chip=2E

On Thu, Jun 6, 2= 024, 10:51=E2=80=AFAM Stuart Cheshire <cheshire@apple=2Ecom> wrote:
On Jun 4, 2024, at 16:03, Rich Brown <richb=2Ehanove= r@gmail=2Ecom> wrote:

> Yeah=2E=2E=2E I didn't write that as carefully as I could have=2E I w= as switching between "user voice" (who'll say 'speed') and "expert" voice (= I know the difference)=2E Check it now: https://randomneuronsfiring=2Ecom/all-the-= reasons-that-bufferbloat-isnt-a-problem/

Thanks for doing that=2E

How about also changing =E2=80=9Cnew faster ISP plan=E2=80=9D to =E2=80=9C= new bigger ISP plan=E2=80=9D? I know that may sound like a slightly weird p= hrase, but getting people=E2=80=99s attention by surprising them a little c= an be beneficial=2E If it looks weird to them and that makes them pause and= think, then that=E2=80=99s good=2E

If the hypothetical ISP imagined here were actually willing to offer a pla= n that truly provided consistently *faster* connectivity instead of just mo= re of the same, we=E2=80=99d be very happy=2E The truth today is that most = IPs offer *bigger*, not *better*=2E They are selling quantity, not quality= =2E

(I am intentionally not lumping *all* ISPs into the same bucket here=2E So= me, like Comcast, are actually making big efforts to improve quality as wel= l as quantity=2E Comcast dramatically reduced the working latency of my cab= le modem during the work-from-home pandemic, and they continue to work on i= mproving that even more=2E I want to be sure to give credit where it is des= erved=2E)

Stuart Cheshire

-- Sent from my Android device with K-9 Mail=2E Please excuse my brevity=2E<= /div>
------5QIWZKRU9V22EXFH1X5H15JGN04ZHV--