Re: [hops] Proposal for HOPS RG

On 22 May 2015, at 10:38, marcelo bagnulo braun wrote:

> I dont think this should be a short lived effort.
> I mean, at the IETF there is plenty of times that we try to design a 
> protocol or protocol extension and we dont really have the data to 
> perform an informed decision.  think the HOPS RG could be a starting 
> point for many protocol design efforts.
> For instance, in the TCPINC BOF, the question about whether encrypted 
> TCP connection would fly over different ports was raised and there was 
> no data available. (and it was a fundamental question to understand if 
> the whole effort was worthwhile)
> Similar questions now are raised in TCPM when designing the extended 
> option format. And again, there is little data around (at least for 
> some aspects of it).

agree.  others I can think of, such as multi path tcp, many rai 
protocols,

> It would be nice to have a place where people that want to work on 
> design can gather data about what works and what doesnt. It would be 
> nice if that was the HOPS RG, i guess.
>
> In other words, one way of doing this is for the HOPS RG t be a venue 
> for people with interesting questions and people who want to measure 
> intersting things (or for people with interesting questions and people 
> who has data that can help them answer the questions)
>
> So, imho, something like HOPS is really missing in the IETF protocol 
> design approach. But maybe it is just me.

not. count me in! deployability has not been so much considered. Many 
protocols have been designed and then a fallback to http/443 was added 
later which makes the protocol brittle.

Marc.

>
>
>
>
> El 22/05/15 a las 16:06, Eggert, Lars escribió:
>> Hi,
>>
>> On 2015-5-22, at 15:46, Mirja Kühlewind 
>> <mirja.kuehlewind@tik.ee.ethz.ch> wrote:
>>> there are people from RIPE who are interested in this work and were 
>>> already at the BarBoF. Further we are also in contact which the 
>>> people from CAIDA. And, as you can see on the agenda, we are also 
>>> talking to Google and Akamai with people who were also at the BarBoF
>> so that's promising, but not actually a large number of folks. I 
>> wonder if a discussion among four groups really needs an RG 
>> established. Isn't this something that might as well be handled  ad 
>> hoc?
>>
>> A second concern I have is that the topic here is fairly narrow in 
>> scope ("let's discuss data around how bad middleboxes break things"), 
>> and rather short-lived (i.e., once that is done, the group is done). 
>> The IRTF tries to charter groups that are long-lived and try to 
>> tackle problem areas of substantial size, and I wonder if this is the 
>> case here.
>>
>> (Since I was not at the bar BOF, I may be fundamentally 
>> misunderstanding something about this proposal. I'm only going on 
>> what is in the charter text proposal.)
>>
>> Lars
>>
>>
>> _______________________________________________
>> hops mailing list
>> hops@ietf.org
>> https://www.ietf.org/mailman/listinfo/hops
>
> _______________________________________________
> hops mailing list
> hops@ietf.org
> https://www.ietf.org/mailman/listinfo/hops