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

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

From: Willy Tarreau <w@1wt.eu>
Date: Tue, 24 Jun 2014 12:20:30 +0200
To: Greg Wilkins <gregw@intalio.com>
Cc: HTTP Working Group <ietf-http-wg@w3.org>
Message-ID: <20140624102030.GC25779@1wt.eu>
Hi Greg,

On Tue, Jun 24, 2014 at 12:07:59PM +0200, Greg Wilkins wrote:
> Willy,
> 
> Do you have a feel for how much processing costs would affect such proxies
> if they still ran with 256kB buffers, but which contained many 16kB frames?
> 
> Is the problem here not so much the frame size, but the proxies requirement
> to decode/inspect frames?
> 
> ie what would be the impost of the proxy read a 256KB buffer, quickly
> scanned it for any new streams, and if none found just forwarded to entire
> 256KB buffer?

That cannot realistically work, because the front LB is a multiplexer in
practice. It receives hundreds of thousands of client connections that it
dispatches to tens of servers, and a single stream from a client is usually
dispatched to multiple servers depending on routing rules (path prefix,
presence of cookies or not, etc).

Instead, (I remember that it was already discussed here in the past), I
really think we'd need to support large frames for large data sets, that
are typically usable for such sites which have few streams per client but
very large ones.

Regards,
Willy
Received on Tuesday, 24 June 2014 10:21:07 UTC

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