Re: HTTP over SCTP without chunked encoding

I'm far from an expert in SCTP, but at first glance, this is  
potentially attractive, but slightly more intrusive in the HTTP stack,  
because it's introducing a new way to frame messages, rather than  
reusing exising code to do so.

The current proposals (AIUI) are less intrusive, because at a certain  
point a SCTP stream just looks like a TCP connection.

Now, whether this trade-off is worth it or not is an open question, of  
course. I think most people are focused on the aspects of SCTP that  
relieve the problems experienced with pipelining, and while having a  
more efficient framing mechanism is interesting, it's not low-hanging  
fruit.

Perhaps this should be an open design question in a HTTP-over-SCTP  
draft, to try to get feedback from interested vendors...

Cheers,



On 02/04/2009, at 11:58 AM, leighton@mail.eecis.udel.edu wrote:

> At the recent HTTP WG meeting the subject of HTTP over SCTP was  
> discussed,
> and in particular the question was raised as to whether or not SCTP  
> could
> be used to avoid chunked encoding.  There are two ways to do this,  
> but one
> of them (EXPLICIT_EOR socket option) conflicts with current thinking  
> about
> how best to send HTTP over multiple SCTP streams.  The other  
> approach is
> to use the payload protocol identified or PPID.
>
> The PPID is an optional value that is set by the sender and read by  
> the
> receiver.  The approach would be to define two PPID values  
> (allocated by
> IANA), say HTTP_MESSAGE_PIECE and HTTP_MESSAGE_END.  The sender  
> would set
> the PPID to HTTP_MESSAGE_END for the last sctp_sendmsg() call for the
> current HTTP object, thereby allowing the receiver to identify the  
> end of
> the current HTTP object.
>
> I'm far from being an expert in HTTP, and I would be grateful for  
> feedback
> on the suitability of this mechanism.  Thanks very much.
>
> - Jon Leighton
>
>


--
Mark Nottingham     http://www.mnot.net/