From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-ob0-x233.google.com (mail-ob0-x233.google.com [IPv6:2607:f8b0:4003:c01::233]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (Client CN "smtp.gmail.com", Issuer "Google Internet Authority G2" (verified OK)) by huchra.bufferbloat.net (Postfix) with ESMTPS id 488CA21F269 for ; Sun, 29 Mar 2015 10:02:36 -0700 (PDT) Received: by obbgh1 with SMTP id gh1so41591908obb.1 for ; Sun, 29 Mar 2015 10:02:36 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:cc :content-type; bh=rcOeNXPkOlrvpAw0z7w6pmkSZ7QIFJCN7oxnATc5+3U=; b=VWCHtfel9XvgTFOhEK6x96hL84cO4RVInoNSf7xJz4EnpmF7HQbYaLoQEImQ+rpMUf KKloj7Y2WCApKCBTFUZo1Zc3aIqnvaufDvu+DhPUgNjVsoHV5AWTHIq+CEm1bPIX4Io6 k4tWyV6gqw6/hb0+GNFqovP+PllKPAh3ICBBV7S604fHZAuEBeTTxNL1MU2qo8HBPfSH r0vkSnvA4OZfAy2oTwVj6oxt5fVBqPzUhiNeNULtXwxEVuRUP5JlicCH2upvp6kH8jRF OIXdPsDLjvEjzStLDwLdIHMUpfSnddRGG3oXCL8KBanT7s77cNXdgVEyhLznUS0t5wMY PURA== MIME-Version: 1.0 X-Received: by 10.202.63.132 with SMTP id m126mr21950634oia.33.1427648556031; Sun, 29 Mar 2015 10:02:36 -0700 (PDT) Received: by 10.202.216.69 with HTTP; Sun, 29 Mar 2015 10:02:35 -0700 (PDT) In-Reply-To: References: <5BD9D41C-9631-4BD0-9C18-7909E9D6E9C0@gmail.com> Date: Sun, 29 Mar 2015 22:32:35 +0530 Message-ID: From: sahil grover Cc: bloat@lists.bufferbloat.net Content-Type: multipart/alternative; boundary=001a113d660a0be30d05127055dc Subject: Re: [Bloat] Requirements for bufferbloat to occur X-BeenThere: bloat@lists.bufferbloat.net X-Mailman-Version: 2.1.13 Precedence: list List-Id: General list for discussing Bufferbloat List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sun, 29 Mar 2015 17:03:05 -0000 --001a113d660a0be30d05127055dc Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: quoted-printable . On Sun, Mar 29, 2015 at 10:04 PM, sahil grover wrote: > Thanks a lot for replying. > > can you please explain me the concept of delay-bandwidth product,window > size and buffer size(or something related with pipe size and packets in > flight). > > because i am unable to understand it from papers/articles etc. > > And the way you explained to me everytime, was very effective. > > So please help in making me understand this(BDP,window size) concept too = . > > > > > > On Sun, Mar 29, 2015 at 2:09 AM, Jonathan Morton > wrote: > >> >> > On 29 Mar, 2015, at 11:04, sahil grover >> wrote: >> > >> > (1) All say bufferSize should be set very large for bufferbloat to >> occur. >> > >> > But how much large?? is there any condition? >> >> There=E2=80=99s a clue in the name: if the buffer is significantly large= r than it >> needs to be (and is unmanaged), we call that bufferbloat. >> >> If the buffer is too small to absorb a typical burst of packets, the >> resulting increase in packet loss will cause a reduction in throughput. >> The correct size for an unmanaged buffer is typically the delay-bandwidt= h >> product, which enables it to absorb a transitory burst from a single TCP >> flow. >> >> However, determining the delay is difficult a priori, and frequently >> differs substantially between different flows on the same connection. S= o >> we usually make some reasonable assumption about the delay component of >> that formula: 100ms is typical for a broadband connection to the public >> Internet, and VoIP can just-about cope with that in practice. >> >> Or, to put it another way - if the buffer *induces* significantly more >> than 100ms delay under load, that is bufferbloat. >> >> > (2) even after setting buffersize very very large, if packets get >> dropped due to buffer >> > overflow when heavy traffic is there. >> > >> > is it bufferbloat? >> >> Yes. Packet loss has nothing to do with it - it=E2=80=99s the induced d= elay that >> matters. >> >> > sholud we take care that maximum limit of buffersize is never reached >> and no >> > packet drop is there due to queue overlow for bufferbloat condition? >> >> Extremely large buffers are usually the result of hardware engineers >> naively attempting to achieve zero packet loss, by providing buffers lar= ger >> than the TCP receive window size. (That is a futile goal - rwnd is >> unlimited in modern operating systems which support window scaling.) >> However, zero packet loss is not a necessary condition. >> >> AQMs often deliberately drop packets in order to signal congestion to th= e >> endpoints. Under some circumstances, this can actually result in less >> overall packet loss than on an unmanaged buffer. Even without ECN, the = AQM >> rarely causes burst losses, whereas overflowing an unmanaged queue often >> does. With ECN, an AQM can often signal congestion sufficiently well >> without dropping any packets at all. >> >> - Jonathan Morton >> >> > --001a113d660a0be30d05127055dc Content-Type: text/html; charset=UTF-8 Content-Transfer-Encoding: quoted-printable
.

On Sun, Mar 29, 2015 at 10:04 PM, sahil grover <= ;sahilgrover0= 13@gmail.com> wrote:
Thanks a lot for replying.

can you please exp= lain me =C2=A0the concept of=C2=A0=C2=A0delay-bandwid= th product,window size and buffer size(or something related with pipe size and packets in flight).=

<= /div>
=C2=A0because i am u= nable to understand it from papers/articles etc.

And the way you explained to me everytime, was = very effective.

So pl= ease help in making me understand this(BDP,window size) concept too .




On Sun, Mar 29, 2015 at 2= :09 AM, Jonathan Morton <chromatix99@gmail.com> wrote:

> On 29 Mar, 2015, at 11:04, sahil grover <sahilgrover013@gmail.com> wrote:=
>
> (1) All say bufferSize should be set very large for bufferbloat to occ= ur.
>
> But how much large?? is there any condition?

There=E2=80=99s a clue in the name: if the buffer is significantly l= arger than it needs to be (and is unmanaged), we call that bufferbloat.

If the buffer is too small to absorb a typical burst of packets, the result= ing increase in packet loss will cause a reduction in throughput.=C2=A0 The= correct size for an unmanaged buffer is typically the delay-bandwidth prod= uct, which enables it to absorb a transitory burst from a single TCP flow.<= br>
However, determining the delay is difficult a priori, and frequently differ= s substantially between different flows on the same connection.=C2=A0 So we= usually make some reasonable assumption about the delay component of that = formula: 100ms is typical for a broadband connection to the public Internet= , and VoIP can just-about cope with that in practice.

Or, to put it another way - if the buffer *induces* significantly more than= 100ms delay under load, that is bufferbloat.

> (2) even after setting buffersize very=C2=A0 very large,=C2=A0 if pack= ets get dropped due to buffer
> overflow=C2=A0 when heavy traffic is there.
>
> is it bufferbloat?

Yes.=C2=A0 Packet loss has nothing to do with it - it=E2=80=99s the = induced delay that matters.

> sholud we take care that maximum limit of buffersize is never reached = and no
> packet=C2=A0 drop is there due to queue overlow for bufferbloat condit= ion?

Extremely large buffers are usually the result of hardware engineers= naively attempting to achieve zero packet loss, by providing buffers large= r than the TCP receive window size.=C2=A0 (That is a futile goal - rwnd is = unlimited in modern operating systems which support window scaling.)=C2=A0 = However, zero packet loss is not a necessary condition.

AQMs often deliberately drop packets in order to signal congestion to the e= ndpoints.=C2=A0 Under some circumstances, this can actually result in less = overall packet loss than on an unmanaged buffer.=C2=A0 Even without ECN, th= e AQM rarely causes burst losses, whereas overflowing an unmanaged queue of= ten does.=C2=A0 With ECN, an AQM can often signal congestion sufficiently w= ell without dropping any packets at all.

=C2=A0- Jonathan Morton



--001a113d660a0be30d05127055dc--