From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from smtp105.iad3a.emailsrvr.com (smtp105.iad3a.emailsrvr.com [173.203.187.105]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (Client did not present a certificate) by huchra.bufferbloat.net (Postfix) with ESMTPS id D890321F1D3 for ; Thu, 15 May 2014 18:01:47 -0700 (PDT) Received: from localhost (localhost.localdomain [127.0.0.1]) by smtp6.relay.iad3a.emailsrvr.com (SMTP Server) with ESMTP id A3A731A80B6; Thu, 15 May 2014 21:01:45 -0400 (EDT) X-Virus-Scanned: OK Received: from app28.wa-webapps.iad3a (relay.iad3a.rsapps.net [172.27.255.110]) by smtp6.relay.iad3a.emailsrvr.com (SMTP Server) with ESMTP id 7D65F1A80B5; Thu, 15 May 2014 21:01:45 -0400 (EDT) Received: from reed.com (localhost.localdomain [127.0.0.1]) by app28.wa-webapps.iad3a (Postfix) with ESMTP id 68D508030C; Thu, 15 May 2014 21:01:45 -0400 (EDT) Received: by apps.rackspace.com (Authenticated sender: dpreed@reed.com, from: dpreed@reed.com) with HTTP; Thu, 15 May 2014 21:01:45 -0400 (EDT) Date: Thu, 15 May 2014 21:01:45 -0400 (EDT) From: dpreed@reed.com To: "David Lang" MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_20140515210145000000_89893" Importance: Normal X-Priority: 3 (Normal) X-Type: html In-Reply-To: References: <3583FA43-EF5B-42D7-A79C-54C87AA514D5@gmail.com> <5373EEFF.5030407@pollere.com> <1400161663.82913275@apps.rackspace.com> Message-ID: <1400202105.4274862@apps.rackspace.com> X-Mailer: webmail7.0 Cc: cerowrt-devel , bloat Subject: Re: [Cerowrt-devel] =?utf-8?q?=5BBloat=5D__fq=5Fcodel_is_two_years_ol?= =?utf-8?q?d?= X-BeenThere: cerowrt-devel@lists.bufferbloat.net X-Mailman-Version: 2.1.13 Precedence: list List-Id: Development issues regarding the cerowrt test router project List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 16 May 2014 01:01:48 -0000 ------=_20140515210145000000_89893 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable =0AI don't think that at all. I suspect you know that. But if I confused yo= u, let me assure you that my view of the proper operating point of the Inte= rnet as a whole is that the expected buffer queue on any switch anywhere in= the Internet is < 1 datagram.=0A =0Afq_codel is a good start, but it still= requires letting buffer queueing increase. However, mathematically, one n= eed not have the queues build up to sustain the control loop that fq_codel = creates.=0A =0AI conjecture that one can create an equally effective conges= tion control mechanism as fq_codel without any standing queues being allowe= d to build up. (Someone should try the exercise of trying to prove that an = optimal end-to-end feedback control system requires queueing delay to be im= posed. I've tried and it's led me to the conjecture that one can always rep= lace a standing queue with a finite memory of past activities - and if one = does, the lack of a standing queue means that the algorithm is better than = any that end up with a standing queue).=0A =0Afq_codel could be redesigned = into a queue-free fq_codel.=0A=0A=0AOn Thursday, May 15, 2014 7:46pm, "Davi= d Lang" said:=0A=0A=0A=0A> If you think "fast lanes" will a= ctually increase performance for any traffic,=0A> you are dreaming.=0A> =0A= > the people looking for "fast lanes" are't trying to get traffic through a= ny=0A> faster, they are looking to charge more for the traffic they are alr= eady=0A> passing.=0A> =0A> David Lang=0A> =0A> On Thu, 15 May 2014, dpree= d@reed.com wrote:=0A> =0A> > Well done. I'm optimistic for deployment ever= ywhere, except CMTS's, the LTE=0A> and HSPA+ access networks, and all corpo= rate firewall and intranet gear.=0A> >=0A> > The solution du jour is to lea= ve bufferbloat in place, while using the real=0A> fads: prioritization (dif= fserv once we have the "fast lanes" fully legal) and=0A> "software defined = networking" appliances that use DPI for packet routing and=0A> traffic mana= gement.=0A> >=0A> > Fixing buffer bloat allows the edge- and enterprise- ne= tworks to be more=0A> efficient, whereas not fixing it lets the edge networ= ks move users up to more and=0A> more expensive "plans" due to frustration = and to sell much more gear into=0A> Enterprises because they are easy marks= for complex gadgets.=0A> >=0A> > But maybe a few engineers who operate and= design gear for such networks will=0A> overcome the incredible business bi= ases against fixing this.=0A> >=0A> > That's why all the efforts you guys h= ave put forth are immensely worth it. I=0A> think this is one of the best = innovations in recent years (Bram Cohen's original=0A> BitTorrent is anothe= r, for fully decentralizing data delivery for the very first=0A> time in a = brilliant way.) I will certainly push everywhere I can to see fq_codel=0A> = deployed.=0A> >=0A> > If there were a prize for brilliant projects, this wo= uld be top on my list.=0A> >=0A> >=0A> >=0A> > On Wednesday, May 14, 2014 9= :25pm, "Dave Taht" =0A> said:=0A> >=0A> >=0A> >=0A> >>= On Wed, May 14, 2014 at 3:32 PM, Kathleen Nichols=0A> =0A> >> wrote:=0A> >> >=0A> >> > Thanks, Rich.=0A> >> >=0A> >> > And to sh= ow you what an amazing bit of work that first fq_codel was,=0A> I have=0A> = >> > on my calendar that I first "exposed" CoDel to a small group in a=0A> = >> > meeting room=0A> >> > and on the phone at ISC on April 24.=0A> >>=0A> = >> And we had all sorts of trouble with the phone, (eric didn't hear=0A> >>= much) and we then spent hours and hours afterwards discussing wifi=0A> >> = instead of codel... it was too much to take in...=0A> >>=0A> >> me, I'd sta= rted jumping up and down in excitement about 20 minutes=0A> >> into kathies= preso...=0A> >>=0A> >> May 3rd, 2012 was the last 24 hr coding stint I thi= nk I'll ever have.=0A> >>=0A> >> https://lists.bufferbloat.net/pipermail/co= del/2012-May/000023.html=0A> >>=0A> >> Ahh, the good ole days, when bufferb= loat was first solved and we all=0A> >> thought the internet would speed up= overnight, and we were going to be=0A> >> rock stars, invited to all the b= est parties, acquire fame and fortune,=0A> >> and be awarded medals and giv= en awards. Re-reading all this brought=0A> >> back memories.... (heck, ther= e's still a couple good ideas in that=0A> >> thread left unimplemented)=0A>= >>=0A> >> https://lists.bufferbloat.net/pipermail/codel/2012-May/thread.ht= ml=0A> >>=0A> >> It looks by may 5th we were getting in shape, and then the= re were a=0A> >> few other issues along the way with the control law and so= on... and=0A> >> eric rewrote the whole thing and made it oodles faster an= d then as=0A> >> best as I recall came up with fq_codel on saturday may 5th= (?) -=0A> >>=0A> >> Ah, I haven't had so much fun in in years. My life sinc= e then seems=0A> >> like an endless string of meetings, politics, and bugfi= xing.=0A> >>=0A> >> The code went from sim/paper, to code, to testing, to m= ainline linux=0A> >> in another week. I wish more research went like that!= =0A> >>=0A> >> commit 76e3cc126bb223013a6b9a0e2a51238d1ef2e409=0A> >> Autho= r: Eric Dumazet =0A> >> Date: Thu May 10 07:51:25 20= 12 +0000=0A> >>=0A> >> codel: Controlled Delay AQM=0A> >>=0A> >> Now, a= s I recall the story, eric came up with fq_codel on a saturday=0A> >> after= noon, so I guess that was may 5th - cinco de mayo!=0A> >>=0A> >> And that t= oo, landed in mainline...=0A> >>=0A> >> commit 4b549a2ef4bef9965d97cbd992ba= 67930cd3e0fe=0A> >> Author: Eric Dumazet =0A> >> Date:= Fri May 11 09:30:50 2012 +0000=0A> >>=0A> >> fq_codel: Fair Queue Co= del AQM=0A> >>=0A> >> let's not forget tom herbert & BQL=0A> >>=0A> >> comm= it 75957ba36c05b979701e9ec64b37819adc12f830=0A> >> Author: Tom Herbert =0A> >> Date: Mon Nov 28 16:32:35 2011 +0000=0A> >>=0A> = >> dql: Dynamic queue limits=0A> >>=0A> >> Implementation of dynami= c queue limits (dql). This is a libary=0A> which=0A> >> allows a queue= limit to be dynamically managed. The goal of dql is=0A> >> to set the= queue limit, number of objects to the queue, to be=0A> minimized=0A> >> = without allowing the queue to be starved.=0A> >>=0A> >>=0A> >>=0A> >>=0A>= >> > It was really amazing to me to watch=0A> >> > something Van and I had= been discussing (okay, arguing) about=0A> privately for=0A> >> > 6 months = and I'd been tinkering with be turned into real code on=0A> real=0A> >> > n= etworks.=0A> >> > Jim Gettys is an incredible (and constructive) nagger, Er= ic Dumazet=0A> and=0A> >> > amazing=0A> >> > coder, and the entire open sou= rce community a really nifty group of=0A> folks.=0A> >> >=0A> >> > Maybe so= meday we will actually update the first article with some of=0A> the=0A> >>= > stuff=0A> >> > we got into the last internet draft....=0A> >> >=0A> >> >= be the change,=0A> >> > Kathie=0A> >> >=0A> >> > O= n 5/14/14 2:01 PM, Rich Brown wrote:=0A> >> >> Folks,=0A> >> >>=0A> >> >> I= just noticed that the announcement for the first testable=0A> >> >> implem= entation of fq_codel was two days ago today - 14 May=0A> 2012.=0A> >> >> Bu= ild 3.3.6-2 was the first to include working fq_codel.=0A> >> >>=0A> >>=0A>= https://lists.bufferbloat.net/pipermail/cerowrt-devel/2012-May/000233.html= =0A> >> >>=0A> >> >> Two years later, we see fq_codel being adopted lots o= f places.=0A> As=0A> >> >> more and more people/organizations come to under= stand the=0A> problem,=0A> >> >> and how straightforward the solution can b= e, we're beginning to=0A> win=0A> >> >> the battle to have a good Internet = experience everywhere.=0A> >> >>=0A> >> >> Thanks to Dave, Eric, Kathie, Va= n, and all the members of this=0A> list=0A> >> >> for their perseverance, t= esting, comments, and patience.=0A> >> >> Congratulations!=0A> >> >>=0A> >>= >> Best regards,=0A> >> >>=0A> >> >> Rich ________________________________= _______________ Bloat=0A> mailing=0A> >> >> list Bloat@lists.bufferbloat.ne= t=0A> >> >> https://lists.bufferbloat.net/listinfo/bloat=0A> >> >>=0A> >> >= =0A> >> > _______________________________________________=0A> >> > Bloat ma= iling list=0A> >> > Bloat@lists.bufferbloat.net=0A> >> > https://lists.buff= erbloat.net/listinfo/bloat=0A> >>=0A> >>=0A> >>=0A> >> --=0A> >> Dave T=C3= =A4ht=0A> >>=0A> >> NSFW:=0A> >>=0A> https://w2.eff.org/Censorship/Internet= _censorship_bills/russell_0296_indecent.article=0A> >> ____________________= ___________________________=0A> >> Cerowrt-devel mailing list=0A> >> Cerowr= t-devel@lists.bufferbloat.net=0A> >> https://lists.bufferbloat.net/listinfo= /cerowrt-devel=0A> >>_______________________________________________=0A> Bl= oat mailing list=0A> Bloat@lists.bufferbloat.net=0A> https://lists.bufferbl= oat.net/listinfo/bloat=0A> ------=_20140515210145000000_89893 Content-Type: text/html; charset="UTF-8" Content-Transfer-Encoding: quoted-printable

I don't th= ink that at all. I suspect you know that. But if I confused you, let me ass= ure you that my view of the proper operating point of the Internet as a who= le is that the expected buffer queue on any switch anywhere in the Internet= is < 1 datagram.

=0A

 

=0Afq_codel is a good start, but it still requi= res letting buffer queueing increase.  However, mathematically, one ne= ed not have the queues build up to sustain the control loop that fq_codel c= reates.

=0A

 

=0A

I conjecture that one can create an equally effective con= gestion control mechanism as fq_codel without any standing queues being all= owed to build up. (Someone should try the exercise of trying to prove that = an optimal end-to-end feedback control system requires queueing delay to be= imposed. I've tried and it's led me to the conjecture that one can always = replace a standing queue with a finite memory of past activities - and if o= ne does, the lack of a standing queue means that the algorithm is better th= an any that end up with a standing queue).

=0A

 

=0A

fq_codel could be rede= signed into a queue-free fq_codel.

=0A

<= br class=3D"WM_COMPOSE_SIGNATURE_END" />

On Thursday, May 15, 20= 14 7:46pm, "David Lang" <david@lang.hm> said:

=0A
=0A

> If you= think "fast lanes" will actually increase performance for any traffic,
> you are dreaming.
>
> the people looking for "fast = lanes" are't trying to get traffic through any
> faster, they are l= ooking to charge more for the traffic they are already
> passing.>
> David Lang
>
> On Thu, 15 May 2014, = dpreed@reed.com wrote:
>
> > Well done. I'm optimistic= for deployment everywhere, except CMTS's, the LTE
> and HSPA+ acce= ss networks, and all corporate firewall and intranet gear.
> >> > The solution du jour is to leave bufferbloat in place, while = using the real
> fads: prioritization (diffserv once we have the "f= ast lanes" fully legal) and
> "software defined networking" applian= ces that use DPI for packet routing and
> traffic management.
= > >
> > Fixing buffer bloat allows the edge- and enterpris= e- networks to be more
> efficient, whereas not fixing it lets the = edge networks move users up to more and
> more expensive "plans" du= e to frustration and to sell much more gear into
> Enterprises beca= use they are easy marks for complex gadgets.
> >
> > = But maybe a few engineers who operate and design gear for such networks wil= l
> overcome the incredible business biases against fixing this.> >
> > That's why all the efforts you guys have put fo= rth are immensely worth it. I
> think this is one of the best inno= vations in recent years (Bram Cohen's original
> BitTorrent is anot= her, for fully decentralizing data delivery for the very first
> ti= me in a brilliant way.) I will certainly push everywhere I can to see fq_co= del
> deployed.
> >
> > If there were a prize= for brilliant projects, this would be top on my list.
> >
= > >
> >
> > On Wednesday, May 14, 2014 9:25pm, = "Dave Taht" <dave.taht@gmail.com>
> said:
> >
> >
> >
> >> On Wed, May 14, 2014 at 3:32 P= M, Kathleen Nichols
> <nichols@pollere.com>
> >>= ; wrote:
> >> >
> >> > Thanks, Rich.
> >> >
> >> > And to show you what an amazing= bit of work that first fq_codel was,
> I have
> >> &= gt; on my calendar that I first "exposed" CoDel to a small group in a
= > >> > meeting room
> >> > and on the phone at= ISC on April 24.
> >>
> >> And we had all sort= s of trouble with the phone, (eric didn't hear
> >> much) and= we then spent hours and hours afterwards discussing wifi
> >>= ; instead of codel... it was too much to take in...
> >>
> >> me, I'd started jumping up and down in excitement about 20 m= inutes
> >> into kathies preso...
> >>
>= ; >> May 3rd, 2012 was the last 24 hr coding stint I think I'll ever = have.
> >>
> >> https://lists.bufferbloat.net/p= ipermail/codel/2012-May/000023.html
> >>
> >> A= hh, the good ole days, when bufferbloat was first solved and we all
&g= t; >> thought the internet would speed up overnight, and we were goin= g to be
> >> rock stars, invited to all the best parties, acq= uire fame and fortune,
> >> and be awarded medals and given a= wards. Re-reading all this brought
> >> back memories.... (he= ck, there's still a couple good ideas in that
> >> thread lef= t unimplemented)
> >>
> >> https://lists.buffer= bloat.net/pipermail/codel/2012-May/thread.html
> >>
>= >> It looks by may 5th we were getting in shape, and then there were= a
> >> few other issues along the way with the control law a= nd so on... and
> >> eric rewrote the whole thing and made it= oodles faster and then as
> >> best as I recall came up with= fq_codel on saturday may 5th(?) -
> >>
> >> Ah= , I haven't had so much fun in in years. My life since then seems
>= >> like an endless string of meetings, politics, and bugfixing.
> >>
> >> The code went from sim/paper, to code, to= testing, to mainline linux
> >> in another week. I wish more= research went like that!
> >>
> >> commit 76e3= cc126bb223013a6b9a0e2a51238d1ef2e409
> >> Author: Eric Dumaze= t <edumazet@google.com>
> >> Date: Thu May 10 07:51:2= 5 2012 +0000
> >>
> >> codel: Controlled De= lay AQM
> >>
> >> Now, as I recall the story, e= ric came up with fq_codel on a saturday
> >> afternoon, so I = guess that was may 5th - cinco de mayo!
> >>
> >&g= t; And that too, landed in mainline...
> >>
> >>= ; commit 4b549a2ef4bef9965d97cbd992ba67930cd3e0fe
> >> Author= : Eric Dumazet <edumazet@google.com>
> >> Date: Fri M= ay 11 09:30:50 2012 +0000
> >>
> >> fq_code= l: Fair Queue Codel AQM
> >>
> >> let's not for= get tom herbert & BQL
> >>
> >> commit 7595= 7ba36c05b979701e9ec64b37819adc12f830
> >> Author: Tom Herbert= <therbert@google.com>
> >> Date: Mon Nov 28 16:32:35= 2011 +0000
> >>
> >> dql: Dynamic queue li= mits
> >>
> >> Implementation of dynamic qu= eue limits (dql). This is a libary
> which
> >> = allows a queue limit to be dynamically managed. The goal of dql is
&g= t; >> to set the queue limit, number of objects to the queue, to = be
> minimized
> >> without allowing the queue to= be starved.
> >>
> >>
> >>
= > >>
> >> > It was really amazing to me to watch<= br />> >> > something Van and I had been discussing (okay, argu= ing) about
> privately for
> >> > 6 months and I'd= been tinkering with be turned into real code on
> real
> &= gt;> > networks.
> >> > Jim Gettys is an incredible = (and constructive) nagger, Eric Dumazet
> and
> >> &g= t; amazing
> >> > coder, and the entire open source commun= ity a really nifty group of
> folks.
> >> >
&= gt; >> > Maybe someday we will actually update the first article w= ith some of
> the
> >> > stuff
> >> = > we got into the last internet draft....
> >> >
&= gt; >> > be the change,
> >> > = Kathie
> >> >
> >> > On 5/14/14 2:0= 1 PM, Rich Brown wrote:
> >> >> Folks,
> >&g= t; >>
> >> >> I just noticed that the announcemen= t for the first testable
> >> >> implementation of fq_c= odel was two days ago today - 14 May
> 2012.
> >> >= ;> Build 3.3.6-2 was the first to include working fq_codel.
> &g= t;> >>
> >>
> https://lists.bufferbloat.net/= pipermail/cerowrt-devel/2012-May/000233.html
> >> >>> >> >> Two years later, we see fq_codel being adopted l= ots of places.
> As
> >> >> more and more peopl= e/organizations come to understand the
> problem,
> >>= ; >> and how straightforward the solution can be, we're beginning to<= br />> win
> >> >> the battle to have a good Interne= t experience everywhere.
> >> >>
> >> >= ;> Thanks to Dave, Eric, Kathie, Van, and all the members of this
&= gt; list
> >> >> for their perseverance, testing, comme= nts, and patience.
> >> >> Congratulations!
> &= gt;> >>
> >> >> Best regards,
> >&g= t; >>
> >> >> Rich ______________________________= _________________ Bloat
> mailing
> >> >> list = Bloat@lists.bufferbloat.net
> >> >> https://lists.buffe= rbloat.net/listinfo/bloat
> >> >>
> >> &g= t;
> >> > _______________________________________________<= br />> >> > Bloat mailing list
> >> > Bloat@li= sts.bufferbloat.net
> >> > https://lists.bufferbloat.net/l= istinfo/bloat
> >>
> >>
> >>
> >> --
> >> Dave T=C3=A4ht
> >>
> >> NSFW:
> >>
> https://w2.eff.org/Censor= ship/Internet_censorship_bills/russell_0296_indecent.article
> >= > _______________________________________________
> >> Cer= owrt-devel mailing list
> >> Cerowrt-devel@lists.bufferbloat.= net
> >> https://lists.bufferbloat.net/listinfo/cerowrt-devel=
> >>_______________________________________________
>= ; Bloat mailing list
> Bloat@lists.bufferbloat.net
> https:= //lists.bufferbloat.net/listinfo/bloat
>

=0A
 ------=_20140515210145000000_89893--