Re: Call for Adoption: TCP Tuning for HTTP

[Willy Tarreau <[e-mail]>]

Hi Mark,

On Wed, Mar 02, 2016 at 04:46:30PM +1100, Mark Nottingham wrote:
> [ copying Alison as our Transport Tech Advisor ]
> 
> Daniel has kindly started a document about how HTTP uses TCP, both for /1 and /2:
>   <https://tools.ietf.org/html/draft-stenberg-httpbis-tcp>
> 
> We haven't explicitly discussed this at a meeting, but I have heard interest in this topic from a variety of folks.
> 
> What do people think about adopting this with a target of Best Current Practice?

I definitely support this. Ideally it would evolve in the direction of
explaining a few shortcomings of TCP when used for HTTP and even suggest
a few implementation hints to prevent newcomers from falling into well
known traps (eg: avoid closing on the client first, or disable lingering
before closing to save the host from ending up with tons of time_wait
sockets that prevent new outgoing connections from being initiated).

I suspect we'll come to a point where people disagree on some settings,
which will be good since tuning is always a tradeoff. It means there
could be proposals for various use cases and their side effects.

I hope that ultimately this document could be referenced in future HTTP
specs so that we don't have to choose between explaining TCP out of
context and dropping the important explanations of some expected trouble.

Best regards,
Willy