On Apr 18, 2014, at 11:43 AM, Thomas Nadeau <[email protected]> wrote:

> 
> On Apr 18, 2014:2:37 PM, at 2:37 PM, Andy Bierman <[email protected]> wrote:
> 
>> 
>> 
>> 
>> On Fri, Apr 18, 2014 at 11:30 AM, Thomas Nadeau <[email protected]> 
>> wrote:
>> 
>> On Apr 18, 2014:2:21 PM, at 2:21 PM, Jamal Hadi Salim <[email protected]> 
>> wrote:
>> 
>> > On Fri, Apr 18, 2014 at 1:39 PM, Thomas Nadeau <[email protected]> 
>> > wrote:
>> >>
>> >>        Why is that not useful? Because I didn't say I was in favor of 
>> >> forces? *)
>> >
>> > Not at all. Your statement was more like a high five. It is ok to
>> > raise your hand at
>> > the meeting - but i was hoping the list would have more useful
>> > technical discussion.
>> >
>> >> It IS useful to say that my company's products implement Yang/Netconf and 
>> >> will support RestConf, >and therefore a derivative of those technologies 
>> >> used for i2rs is preferred. We have no plans to >support any of the other 
>> >> discussed options in i2rs such as Forces at this time.  The reasons are 
>> >> >simple: our customers asked for Netconf/Yang models and as far as I can 
>> >> tell, have no interest in >Forces.
>> >>
>> >
>> > So again it boils down to "bussiness" reasons - am i wrong?
>> 
>>         Yes, but not in the way you imply. My company has implemented 
>> Yang/Netconf because our customers have asked us to and because when we did, 
>> they bought our equipment - not as you imply, because we think its cool. 
>> There are a number of other major hardware vendors on this list that have 
>> done the same. I think if you poll most of those vendors, you will find that 
>> they generally don't  engineering resources to things that people don't buy. 
>>  Also, Open Daylight has gone with Netconf/Yang/RestConf as one of its 
>> primary interfaces - there are dozens of vendors working on that project 
>> too.  As far as I can tell, no one has done any programmable interface based 
>> on Forces there.  That seems to be an overwhelming number of implementations 
>> that have gone with yang/netconf/restconf.
>> 
>> 
>> It would be just as valid for people to chime in "We already use ForCES"
>> or "we use already OF-Config", as a reason to use it for I2RS.
> 
>       It would, and I think that is what the chairs are trying to assess so 
> chime on in! 

 While we (the ONF contributors) have explicitly tried to keep the OF-CONFIG 
specification language and protocol agnostic, but the only currently available 
implementation-ready language and transport mapping available is YANG and 
NETCONF. So in reality, OF-CONFIG is currently NETCONF with a YANG module 
maintained by the ONF.

>       --Tom
> 
>> 
>> 
>> Andy
>> 
>> > "we have an implementation of Yang/netconf";
>> > "we are going to make it work for I2RS".
>> > If thats the case - then lets just end the discussion.
>> > I am hoping to get the answer to:
>> > What are the technical requirements/arguements?
>> > Why do i have to struggle so much in a technical environment like IETF for
>> > someone to come out and say what the requirements are?
>> >
>> > I can assure you that i have no religious conviction that say it has
>> > to be  ForCES.
>> > Lets actually start with you showing me how netconf/yang meets the 
>> > requirements.
>> > Otherwise - please stop with the high fives.
>> >
>> >>        From my perspective, continuing to debate which protocol is 
>> >> derived to become i2rs is a waste >of everyone's time.
>> >
>> > Whats the point to having a circus show in pretending we are trying to
>> > get any consensus?
>> 
>>         I am not WG chair, so I will let Ed/Jeff comment on the circus. *)
>> 
>> > Its clear what people want from the meeting, and if you poll interested 
>> > implementations you'll likely >get the same answer: yang/netconf/restconf. 
>> >  Lets please stop delaying i2rs and get some work >done here.
>> >>
>> >
>> > When you go and ask people to raise their hands they will. You need to
>> > ask them why
>> > they are raising their hands. I am hoping this is not a staked
>> > popularity contest.
>> 
>>         Well, at some point it does come down to running code and real, 
>> deployed implementations, right?  As I said above, there are many real and 
>> significant implementations of yang/netconf in the marketplace *today*. 
>> Asking those things to be modified slightly isn't that big of a deal in 
>> order to address the requirements. Implementing a completely new protocol 
>> is. I am not totally against new protocols - its just that if we can achieve 
>> the goals we have by hacking an existing implementation, that makes better 
>> engineering and business sense.
>> 
>>         --Tom
>> 
>> 
>> 
>> >
>> >
>> > cheers,
>> > jamal
>> >
>> > _______________________________________________
>> > i2rs mailing list
>> > [email protected]
>> > https://www.ietf.org/mailman/listinfo/i2rs
>> >
>> 
>> 
>> _______________________________________________
>> i2rs mailing list
>> [email protected]
>> https://www.ietf.org/mailman/listinfo/i2rs
>> 
>> 
> 
> _______________________________________________
> i2rs mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/i2rs

--
Carl Moberg
VP Technology
twitter: @cmoberg
http://www.tail-f.com/

_______________________________________________
i2rs mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/i2rs

Reply via email to