- From: Willy Tarreau <w@1wt.eu>
- Date: Fri, 18 Jan 2013 15:11:30 +0100
- To: RUELLAN Herve <Herve.Ruellan@crf.canon.fr>
- Cc: Mark Nottingham <mnot@mnot.net>, Martin J. Dürst <duerst@it.aoyama.ac.jp>, Roberto Peon <grmocg@gmail.com>, Nico Williams <nico@cryptonector.com>, James M Snell <jasnell@gmail.com>, "ietf-http-wg@w3.org" <ietf-http-wg@w3.org>
Hi Hervé, On Fri, Jan 18, 2013 at 01:58:12PM +0000, RUELLAN Herve wrote: > I'll try a shot at the URLs. Experimental data show that URLs often share the same beginning: for requests targeting a web sites, the URLs will usually start with the same scheme and host and possibly port. The beginning of the path is also usually shared by several URLs. > > Therefore an efficient encoding for an URL is as a delta from a previous URL: the number of shared characters at the beginning, and the new characters. To reduce the state that need to be stored, it is possible to use only the previous URL as a reference. Some testing was already done with this several months ago and showed promising results. However, I didn't have enough captures to play with and I recall that Roberto had was more reserved on whether this would work all the time or not. I must admit that another issue concerns the amount of state to be kept between two requests on intermediaries. Some unusually long URIs can be several kB large, which would cause a real pain if they have to persist during idle time. Regards, Willy
Received on Friday, 18 January 2013 14:12:15 UTC