W3C home > Mailing lists > Public > ietf-http-wg@w3.org > January to March 2016

Re: [tcpm] Call for Adoption: TCP Tuning for HTTP

From: Ben Niven-Jenkins <ben@niven-jenkins.co.uk>
Date: Sat, 5 Mar 2016 12:54:08 +0000
Cc: Tim Wicinski <tjw.ietf@gmail.com>, Mark Nottingham <mnot@mnot.net>, Joe Touch <touch@ISI.EDU>, Patrick McManus <mcmanus@ducksong.com>, HTTP Working Group <ietf-http-wg@w3.org>
Message-Id: <38CB1EBC-3C38-430A-998A-9AF426A2AA99@niven-jenkins.co.uk>
To: Willy Tarreau <w@1wt.eu>
Willy, 

> On 5 Mar 2016, at 07:38, Willy Tarreau <w@1wt.eu> wrote:
> 
>> On Fri, Mar 04, 2016 at 08:05:31PM -0500, Tim Wicinski wrote:
>> 
>> DNSOP recently went through the process of updating RFC5966,
>>    "DNS Transport over TCP - Implementation Requirements"
>> 
>> and it was published just this week as RFC 7766
>>    https://datatracker.ietf.org/doc/rfc7766/
>> 
>> There was a lot of effort put in this document to reflect current 
>> technology in TCP, like Fast Open.
> 
> This is a very interesting document, thanks for sharing. This is the
> type of work we hope to see for HTTP, explaining the tradeoffs of
> various solutions.
> 
>> I support adoption, but I also think the document needs work.  I'll be 
>> happy to assist and make it look less like a list of linux kernel tunings,
> 
> The tuning is an important point, but only once all the explanation are
> given for the trouble the tuning aims to fix. And ultimately it should
> cover multiple operating systems.

Does the doc need to cover operating systems at all?

Is it sufficient to have the explanations and describe what to tune and leave it to the reader to work out how to set tunable foo on their operating system?

Ben

> 
> Regards,
> Willy
> 
Received on Saturday, 5 March 2016 12:54:45 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 22 March 2016 12:47:11 UTC