- From: Nicolas Mailhot <nicolas.mailhot@laposte.net>
- Date: Fri, 15 Nov 2013 21:06:32 +0100
- To: "Roberto Peon" <grmocg@gmail.com>
- Cc: "Nicolas Mailhot" <nicolas.mailhot@laposte.net>, "Bruce Perens" <bruce@perens.com>, "Ryan Hamilton" <rch@google.com>, "David Morris" <dwm@xpasc.com>, "James Snell" <jasnell@gmail.com>, "HTTP Working Group" <ietf-http-wg@w3.org>, "Julian Reschke" <julian.reschke@gmx.de>, "Stephen Farrell" <stephen.farrell@cs.tcd.ie>
Le Ven 15 novembre 2013 18:31, Roberto Peon a écrit : > That leaves us with either using a new port (infeasible, failure rate > still > in high 10%s) or doing something else so as to be able to deploy. > What is the something else would you suggest? I'm not convinced at all using a new port is infeasible. Yes, it would be devilishly hard for a new corner-case protocol. But the web as a whole is something else entirely and a new way to access it won't be dismissed so easily. Of course adoption would take years, you don't replace billions of web equipments that easily, but it would happen anyway. That is, provided the protocol actually brings enough on the table to justify the pain. For example bittorent provided value for pirates and Linux users and they authorised in on their equipments. It didn't provide value to anyone else and was blocked elsewhere. Websocket, I'm sorry to say, does not bring any value. Its main purpose is to help developers force the passage of protocols people already blocked, so of course people will block websockets too (notwithstanding what websocket developers wish) So what does the http/2 draft propose today? 1. better latency: that will give you mobile users, gamers and traders (through web services) 2. encryption: I'm sorry that's an antifeature for all average users. Encryption is a PITA to use and deploy, it kills caching, it kills interop, it doesn't remove monitoring by web giants or the NSA, CA security is questioned, people by and large do not care if most of their traffic is observed, and you're proposing tanks to access cardboard houses (most web sites are not ironclad and won't be in the future). Case in point: ipsec market failure, social network successes. 3. server push: again, an antifeature. People are already bombarded by advertisements and other stuff they didn't request. 4. new web applications? Who cares, they don't exist now. By all means enable them but the only thing that will sell your protocol is making better what people already know and want And that's all. That's pretty thin to justify the pain if you're not a mobile operator or a privacy zealot. In fact I suspect it's a net loss so far. So I'd suggest you need to work more on your sales argument instead of trying to force people to adopt something they don't need through technical trickery. It's not as if there are no easy wins out there: 1. caching: http/1 caching is byzantine and not that efficient. You have ideas on the subject? Wonderful! Everyone loves smaller download times and less bandwidth use. 2. compression: surely in the past decade the state of the art improved enough you can add better compression engines as option in http/2? Terrific! again, everyone loves less bandwidth use (but make it possible to optionnaly inhibit it for people with CPU constrains) 3. cookies: magnificent! the wg didn't want to discuss this but EU legislation shows demand exists to fix the cookie hell, and web sites operators would rather pass their time elsewhere than operating cookie banners. 4. intermediaries: fix intermediary deployment problems (in particular paygates) and you fix the hotel/airport/metro station/convention/guest user situation. Extraordinary! Many people would kill to avoid paying xG bills when data networks are available, and resort operators only want to keep away external network leeches. And there are probably others I forget. Remember, web infrastructure as a whole represents an heavy investment by many actors. http has been static for a decade and that's a net plus for most users. Everyone but developers hates technical change. You need to put a lot on the table to convince network operators to let you pass. And you need to present it in a single package "please pay for new equipments and BTW I'll introduce another change next year so be prepared for another bill" is *not* a seller. So putting away problems to look at in another revision is not the way to go. Regards, -- Nicolas Mailhot
Received on Friday, 15 November 2013 20:07:04 UTC