Re: [Taps] First draft of an article for the IETF journal

On 21 May 2014, at 15:41, Aaron Falk <falk-ietf@dgftech.com> wrote:

> On Wed, May 21, 2014 at 5:11 AM, Michael Welzl <michawe@ifi.uio.no> wrote:
> 
> My opinion: at first, we should focus on things that can be installed one-sided, on the sender side only. Yes it's limited, but a number of things are possible there. E.g., pluggable congestion controls in TCP have always worked without involving the receiver - but right now, pluggable congestion control isn't an IETF matter. Happy Eyeballs can be done without coordinating with a receiver, at least for the case that we're using well-known ports and the receiver listens on more than one protocol.
> 
> I've forgotten the details of 'Happy Eyeballs'.  Can you provide a cite?
>  

Hi,

http://tools.ietf.org/html/draft-wing-tsvwg-happy-eyeballs-sctp-02

Cheers,

David

> 
> But yes, quite quickly one runs into limitations there; anyway, for divide-and-conquer  :-)   reasons, I think we want to go with a sender-side-only-installation assumption for now.
> 
> Seems worth exploring the implications of such an approach where you have the same app sitting on top of very different APIs (e.g., Berkeley socket vs TAPS) at either end of the connection.  That combined with an app where either endpoint might be a sender.
> 
>  
> Though, I'm not sure if we have had enough discussions on this and if absolutely everyone is on the same page (though, indirectly perhaps, as our draft charter states that signaling is out of scope, and signaling would be needed to align both ends).
> 
> yup
> 
> cheers,
> 
> --aaron
> _______________________________________________
> Taps mailing list
> Taps@ietf.org
> https://www.ietf.org/mailman/listinfo/taps