Re: [grobj] Referrals problem statement at IETF 79

[Sheng Jiang <[e-mail]>]

In 2010/11/9 18:07, Brian E Carpenter wrote:
> I presented the draft Problem Statement for Referral
> <draft-carpenter-referral-ps-01.txt>  by myself, Sheng Jiang, and
> Bo Zhou (departing author) and Zhen Cao (new author, ChinaMobile),
> three times: in the two open meetings for the Applications and
> Transport Areas, and in the Name Based Sockets (NBS) BOF.

Thanks Brian for the hard work.

> There was significant interest and feedback in the APPAREA meeting,
> with comments that we had identified an important problem but perhaps
> not the correct direction for a solution. There was feedback that
> applications people also want a solution to the path selection problem
> (in other words, the network layer should fix the mess it's made over
> the last 15 years with NATs and two versions of IP).

I guess we are in very early stage of designing the solution. We would 
like to hear any suggestion. We are welcome to the requirements from 
application side. I guess only after we clearly describe the problems we 
need to solve, we can decide with direction we may go.

> In the TSVAREA meeting, there was some positive discussion and agreement
> that the problem statement should be further developed.

We are working on the problem statement, for sure. But at the same time, 
we (or IETF) should decide which area this work should be hold. 
Personally, I think this is work in transport layer which serves 
applications eventually. So, it should stay in transport area, but 
requiring application people to be deeply involved.

Regards,

Sheng

> There was less feedback in the NBS BOF. It seems to me that the NBS
> proposers are understimating the issues that they face once referral
> and separate addressing scopes are present. But there is a similar
> message that connectivity should be provided by some code under the
> application layer, not by heuristics in every separate application.
>
> I hope that some people from those three audiences will join
> the discussion of the problem statement here.
>