W3C home > Mailing lists > Public > ietf-http-wg@w3.org > April to June 2014

Re: Stuck in a train -- reading HTTP/2 draft.

From: Johnny Graettinger <jgraettinger@chromium.org>
Date: Thu, 26 Jun 2014 13:25:55 -0400
Message-ID: <CAEn92Tp8ubAZGTqRaLvDGpva6uAziJU70L5F5GDU9krKGpnWVQ@mail.gmail.com>
To: Willy Tarreau <w@1wt.eu>
Cc: Roberto Peon <grmocg@gmail.com>, Poul-Henning Kamp <phk@phk.freebsd.dk>, Martin Thomson <martin.thomson@gmail.com>, Jason Greene <jason.greene@redhat.com>, Mark Nottingham <mnot@mnot.net>, HTTP Working Group <ietf-http-wg@w3.org>
> > This seems a whole lot simpler, allows the load balancer to splice(), and
> > works with SSL.
> Sure but this is layer3-based routing, not l7-based as customers want
> today.

Well, technically SNI is L5, right?

I can't speak for others, but know that in my own experience and having
deployed HAProxy at pretty significant scale in the past (and thank you,
for a great product), we opted for L3 routing anyway.
This was just because it provided service isolation and better matched the
organizational structure of the teams maintaining different application
types. Performance wasn't a deep consideration.

I'm not dismissive of the assertion that customers want L7 routing of mixed
service types with wildly different response properties and performance
requirements, but I personally haven't seen it. And I'd probably try to
talk them out of it :)

And HTTP/2 will make this even more important because by avoiding a new
> connection to retrieve videos, we'll save one RTT and videos will start
> faster.

This can be overcome (preconnect / prefetch hints, etc).
Received on Thursday, 26 June 2014 17:26:23 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:14:31 UTC