Re: [tcpm] [tsvwg] Agenda requests for TSVWG@IETF101

Just to confirm,

On Tue, Mar 13, 2018 at 6:51 AM, Gorry Fairhurst <gorry@erg.abdn.ac.uk>
wrote:

> On 13/03/2018, 11:23, Scharf, Michael (Nokia - DE/Stuttgart) wrote:
>
>> As for TSVWG, we have had quite a long discussion with the authors at the
>>> meeting and after. The TSV chairs encouraged them to take the CC aspects
>>> separately and explain why this method is better and detail what this
>>> benefit
>>> is and what is required in the router to allow this. We suggested an
>>> initial talk
>>> in ICCRG to present results and show *why* this is attractive. As far as
>>> I know
>>> they requested time to do this.
>>>
>> ICCRG is the right home for this. And the TCPM charter explicitly allows
>> us to move topics there...
>>
>> I have written a PhD thesis on exactly this topic; we had a network
>> processor implementation of Quick-Start, and our use cases was actually
>> augmented/virtual reality... And I run into tons of fundamental issues 10
>> years ago, which are not easy to solve and not even mentioned in the I-Ds.
>> So I somehow feel qualified to give feedback on what they have to look at
>> more in detail 😉
>>
>> They also requested time in TSVWG - but there's (as yet) been little
>>> discssion
>>> on the list, so we curently advise them to prepare a slide to show to
>>> say why
>>> people should read the draft. We have not decided (yet) to give time to
>>> this
>>> new framework.
>>>
>> I personally believe that for new ideas the IETF should give a
>> presentation slot *once*. For a second presentation the bar should be
>> higher. If the framework draft has been presented already in TSVWG, IMHO
>> further discussion on that belongs now on the list.
>>
>> They have now also requested time in TCPM.
>>>
>> The draft is very specific about TCP congestion control, so in principle
>> it belongs into TCPM. In general, I think TCPM should be open to new ideas.
>> But my own thinking is that this can be at best a "if time permits"
>> presentation in TCPM.
>>
>> I don't know whethere this time they are also requesting slots some other
>>> places to discuss other aspects.
>>>
>>> I also suggested (informally) that they should try making a great short
>>> presentation of how this is a new opportunity and whar has changed since
>>> we last saw schemes proposed. I do see that there are significant
>>> advances in
>>> router forwarding hardware - and I suspect there could be similar ideas
>>> in
>>> cisco, etc.Would other vendors (or operators) be interested in
>>> standardising
>>> this? I think such a talk could be put to TSVAREA.
>>>
>> I am only at the IETF on Sunday evening and Monday morning.
>> Unfortunately, I may not be in TSVAREA as I have to go back to the airport.
>>
>> I doubt that other people who are familiar with RTG and OPS area will
>> show up in TSVAREA (but I may be wrong). It is pointless to discuss traffic
>> engineering and OAM in router hardware without RTG area input. In RTG and
>> OPS area, the level of multi-vendor support of a document can typically be
>> seen in the author list.
>>
>> Michael
>>
>> I suspect it is far too late to use TSVAREA for this at the coming IETF.
>

https://datatracker.ietf.org/meeting/101/materials/agenda-101-tsvarea looks
pretty packed to me - it's a one-hour slot.

> Actually, to be more clear --- I am **NOT** against some sort of evolution
> in the network, and figuring out how best transports interact with this.
>

Ditto.

Spencer

> I still have people working on this, I see newer router designs as more
> inviting, I still see appetite to do this. But like you, I **KNOW** there
> some big obtsacles and pitfalls. I am wondering how best to bring this sort
> of work to the IETF and get good perspective on the problems and solutions.
> After PLUS was short down by Privacy, I'd hate the next new thing to be
> shot down for the wrong reason.
>
> Gorry
>
>
> _______________________________________________
> tcpm mailing list
> tcpm@ietf.org
> https://www.ietf.org/mailman/listinfo/tcpm
>