W3C home > Mailing lists > Public > ietf-http-wg@w3.org > January to March 2009

Re: Pipelining in HTTP 1.1

From: Roy T. Fielding <fielding@gbiv.com>
Date: Mon, 30 Mar 2009 17:43:10 -0700
Message-Id: <1A7D50CF-497B-427A-84DA-CA841EC3F6F9@gbiv.com>
Cc: "Preethi Natarajan (prenatar)" <prenatar@cisco.com>, ietf-http-wg@w3.org, Jonathan Leighton <leighton@cis.udel.edu>, "Paul D. Amer" <amer@cis.udel.edu>
To: Fred Baker <fred@cisco.com>
On Mar 30, 2009, at 5:25 PM, Fred Baker wrote:
> On Mar 30, 2009, at 5:05 PM, Roy T. Fielding wrote:
>
>> I am somewhat surprised that your HTTP over SCTP spec allows  
>> channels to be reused.  Is there a significant cost associated  
>> with channel allocation?
>
> At session startup, each end tells the other how many streams it  
> wants to open, and either they agree on the lower number or one of  
> them aborts the connection.
>
> That's an argument for a Really Big Number :-)

I thought that they would only cause a limited pool size
(max open channels), but I see now that

RFC4960, sec 5.1.1:

    After the association is initialized, the valid outbound stream
    identifier range for either endpoint shall be 0 to min(local OS,
    remote MIS)-1.

effectively requires the stream identifiers to be reused.  Doh!

....Roy
Received on Tuesday, 31 March 2009 00:43:49 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 27 April 2012 06:51:01 GMT