- From: Roberto Peon <grmocg@gmail.com>
- Date: Tue, 1 Jul 2014 23:39:12 -0700
- To: Poul-Henning Kamp <phk@phk.freebsd.dk>
- Cc: Kazu Yamamoto <kazu@iij.ad.jp>, HTTP Working Group <ietf-http-wg@w3.org>
Received on Wednesday, 2 July 2014 06:39:39 UTC
Poul-- That is orthogonal-- small data frames are necessary for latency as well, and the impact on packetization with full DATA frames is minimal, with near zero impact on latency. -=R On Tue, Jul 1, 2014 at 11:34 PM, Poul-Henning Kamp <phk@phk.freebsd.dk> wrote: > In message < > CAP+FsNc+xW1gKma0McrgXtPpwR0BCubHkvHhUbcHHyn1Sd6t0g@mail.gmail.com>, > Roberto Peon writes: > > >If the headers were regularized, with the use of a reference set, one > could > >imagine a reduction of 20 bytes per header. > >With 100 elements, this is approximately 2k of data, or two packets worth. > > You'd save as many if not more bytes by being able to transfer the > biggest of the 100 elements in DATA frames larger than 16383 bytes. > > I think it is fair to ask for some credible real-world data that shows > reference sets making an actual difference that justifies the added > complexity. > > -- > Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 > phk@FreeBSD.ORG | TCP/IP since RFC 956 > FreeBSD committer | BSD since 4.3-tahoe > Never attribute to malice what can adequately be explained by incompetence. >
Received on Wednesday, 2 July 2014 06:39:39 UTC