W3C home > Mailing lists > Public > ietf-http-wg@w3.org > January to March 2013

Re: Framing and control-frame continuations

From: Brian Pane <brianp@brianp.net>
Date: Wed, 6 Feb 2013 07:13:54 -0800
Message-ID: <CAAbTgTtC1sZD5fasfB_pLTZz=0_C09Kv44hFvRaQXDs+NMQRSA@mail.gmail.com>
To: ietf-http-wg@w3.org
On Wed, Feb 6, 2013 at 6:28 AM, Poul-Henning Kamp <phk@phk.freebsd.dk> wrote:
> Content-Type: text/plain; charset=ISO-8859-1
> --------
>
>
>>The window size needs to be larger than 64k to fully utilize such links.
>
> Just got off the phone, it was the 1980'ies which wanted their
> protocol design parameters back, will you handle that ?  :-)
>
> There is no need to negotiate a maximum frame size, because the
> length comes up front, so the receiver can always chop it up to
> more manageable bits if it cannot process it in the full size.

The "chop it up to more manageable bits" approach works for data
frames, where a load balancer receiving a large frame can chop it up,
thereby maintaining a limit on per-connection memory usage.  But it
doesn't work for control frames; there already are use cases where a
load balancer will need to see the entirety of a control frame in
order to decide where/whether to route it.

Brian
Received on Wednesday, 6 February 2013 15:14:22 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 6 February 2013 15:14:24 GMT