Hi Aleks, On Thu, Apr 07, 2016 at 08:41:58PM +0200, Aleksandar Lazic wrote: > Good to here that there are some plans for H2.
You know, at least for having participated to it, exposed and defended our users' constraints and use cases, and our technical difficulties, even only for this I want to have it. Just as many such wide-scope protocols, it is not perfect but when you consider that it ultimately managed to satisfy all participants from IoT to CDN passing by proxies, browsers and printers, I guess we can safely say that it deserves some attention. > As we see on the answers on the list that there is a demand but not to much > pressure. Yes, that matches our initial estimates, that's pretty encouraging! > It's also a political thing like 'what you can't handle HTTP/2 this > fantastic new solve everything Protocol 8-O', you know. Well not only. There are really good things in HTTP/2. We'll very likely save TLS resources compared to HTTPS for example, because browsers will establish a single TLS connection to haproxy and multiplex all their requests in it, while today they're establishing multiple connections to fetch objects in parallel and these connections remain alive. That will lower the connection count and handshake-to-resume ratio, even in active- active setups since most of the time there will be a single connection per visitor. And pages really load faster on high latency links. So it's normal that people want it. People will start to hurry when it becomes a competition problem. It's not much deployed yet but is showing a steady growth. At some point we'll have to be ready! > Due to the fact that I'm long enough part of the haproxy community I'll keep > an eye on this and will ask again in some time and some times the thread > support is just here ;-). Thanks :-) > If you need some help just give me a ping. OK thanks, willy
