Re: [tcpm] TCP tuning

[Michael Welzl <michawe at ifi.uio.no>]

On Feb 3, 2010, at 6:48 PM, Adam Langley wrote:

> On Wed, Feb 3, 2010 at 12:43 PM, Joe Touch <touch at isi.edu> wrote:
> > IMO, I would encourage Google to take their work there and get  
> > feedback
> > there first. TCPM is a place where mature, vetted ideas should be
> > considered, preferably (IMO).
>
> I think an important part of the presentation is that this is already
> a very well vetted idea: by using multiple subdomains and concurrent
> connections, HTTP has already increased the initcwnd drastically.

I would disagree about "well vetted" (at the level that TCPM
needs to incorporate such a major change in the protocol)
unless I see some very thorough measurements showing
that no harm comes from such a change, in various environments,
with various possible traffic occurrences coming and going...

Talking about this in ICCRG wouldn't be a bad thing at all,
it's really a reasonable first step IMO.

Cheers,
Michael