Resent-Date: Tue, 28 May 2013 20:34:13 +0000
Resent-Message-Id: <E1UhQaz-0002ad-2r@frink.w3.org>
MIME-Version: 1.0
Sender: patrick.ducksong@gmail.com
In-Reply-To: <CAP+FsNez763nkt5EPo8Wf496gH-+hY_V1NRuT5TDuM+697L6_g@mail.gmail.com>
References: <CAOdDvNoAjiRSBv9ue6RgCQJ4wMNQcKBH2a8zVa4_96wbp=g8MA@mail.gmail.com>
 <CABP7Rbefh0HxT7Pui_F8viNvu8232O3Qt=VaR6SgsL1DQarVSA@mail.gmail.com>
 <CAA4WUYgKsDudsSAywWSwz5KVsEV5iUREqjmYVB5sWuc+11ujOQ@mail.gmail.com>
 <CAP+FsNdejY=K4fp6jMh1AzSkMpdxWNd+cCnaF6uw2GPfMVtjAA@mail.gmail.com>
 <CABP7Rbf6Ls8pBf9Rons9hgLeXjnm-yk6t6kebk1EXcS3bTdf_Q@mail.gmail.com>
 <CAA4WUYjGk5EYeP9pP=TDWdGGyq5PjwHcDc+qD1mBGuSAt9yvng@mail.gmail.com>
 <CAP+FsNez763nkt5EPo8Wf496gH-+hY_V1NRuT5TDuM+697L6_g@mail.gmail.com>
Date: Tue, 28 May 2013 16:33:29 -0400
Message-ID: <CAOdDvNr81+HtAQ1bR5LtXHmD1Gz+XbfHP=9WZiNZvHodxUddpQ@mail.gmail.com>
From: Patrick McManus <mcmanus@ducksong.com>
To: Roberto Peon <grmocg@gmail.com>
Cc: =?UTF-8?B?V2lsbGlhbSBDaGFuICjpmYjmmbrmmIwp?= <willchan@chromium.org>,
 James M Snell <jasnell@gmail.com>, HTTP Working Group <ietf-http-wg@w3.org>
Content-Type: multipart/alternative; boundary=001a11c2f0688fa2ce04ddcd2dd7
Received-SPF: pass client-ip=209.85.214.171;
 envelope-from=patrick.ducksong@gmail.com; helo=mail-ob0-f171.google.com
Subject: Re: Proposal - Reduce HTTP2 frame length from 16 to 12 bits
Archived-At: <http://www.w3.org/mid/CAOdDvNr81+HtAQ1bR5LtXHmD1Gz+XbfHP=9WZiNZvHodxUddpQ@mail.gmail.com>
Resent-From: ietf-http-wg@w3.org
Resent-Sender: ietf-http-wg-request@w3.org
Precedence: list

--001a11c2f0688fa2ce04ddcd2dd7
Content-Type: text/plain; charset=ISO-8859-1

On Tue, May 28, 2013 at 4:12 PM, Roberto Peon <grmocg@gmail.com> wrote:

My sweet-spot number was 16k, as I knew that I could saturate a 10G nic
> with 16k frames/writes and have enough CPU left over to do some actual
> work. The amount of overhead goes up more than linearly with the decrease
> in frame size thanks to contention, etc.
>
>

Given what you've said here and in the other mail (plus of course my own
previously stated concerns) I'm inclined to suggest a 16KB max (14 bits)
without introducing any kind of max frame size configurable. My point is to
drive it as small as we can without creating excessive overhead and you've
put a stake in the ground that 16KB is that level. That's still 4x as
aggressive as the current draft.

--001a11c2f0688fa2ce04ddcd2dd7
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><br><div class=3D"gmail_extra"><br><br><div class=3D"gmail=
_quote">On Tue, May 28, 2013 at 4:12 PM, Roberto Peon <span dir=3D"ltr">&lt=
;<a href=3D"mailto:grmocg@gmail.com" target=3D"_blank">grmocg@gmail.com</a>=
&gt;</span> wrote:<br>
<br><blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-lef=
t:1px #ccc solid;padding-left:1ex"><div dir=3D"ltr"><div class=3D"gmail_ext=
ra"><div class=3D"gmail_quote"><div>My sweet-spot number was 16k, as I knew=
 that I could saturate a 10G nic with 16k frames/writes and have enough CPU=
 left over to do some actual work. The amount of overhead goes up more than=
 linearly with the decrease in frame size thanks to contention, etc.</div>

<div>=A0<br></div></div></div></div></blockquote><div><br></div><div>Given =
what you&#39;ve said here and in the other mail (plus of course my own prev=
iously stated concerns) I&#39;m inclined to suggest a 16KB max (14 bits) wi=
thout introducing any kind of max frame size configurable. My point is to d=
rive it as small as we can without creating excessive overhead and you&#39;=
ve put a stake in the ground that 16KB is that level. That&#39;s still 4x a=
s aggressive as the current draft.<br>
</div><div>=A0</div></div></div></div>

--001a11c2f0688fa2ce04ddcd2dd7--