Le 2012-04-09 à 12:16, Maoke a écrit :

> 
> 
> 2012/4/9 Liubing (Leo) <[email protected]>
> >fine. we'd love to see it out with good amount of data and good amount of 
> >feedback from testing operators.
> 
> >if you invite me to test, i'd love to provide challenging scenarios. ;-)
> 
> >but it doesn't mean we should postpone current process of standardization of 
> >MAP, because the result of 4rd-u is still unsure. - maoke 
> 
>  
> I’d love to see operators find proper mechanisms respectively according to 
> their own situations, rather than some mechanisms *out*.
> 
> 
> joke. operators haven't know what the 4rd-u is.

This is disparaging for operator engineers who took the time to read the 
specification (not to speak about those who co-autrhored the specification!).


> operators never make selection on things not belonging to its knowledge. come 
> on, believe me, as an engineer in an operator company, the real operators are 
> more conservative than me. ;-) 
>  
>  
> The transition mechanisms are developed for the operators, but in my 
> observation, the MAP hasn’t  proved sufficiently for operators’ long-term 
> selection. So why we have to rush to a *stand track*?
> 
> 
> it is only your observation and it is a different question.

> you may argue that the MAP is not mature enough for the wg acceptance but the 
> wg is doing a consensus decision.

Precisely, and people who attended MAP and 4rd presentations couldn't reach 
consensus.

> on the other hand, as i pointed out,

> even there is no MAP at all, i don't think 4rd-u now is at the time of 
> talking about wg acceptance. 

WKG acceptance of 4rd as experimental isn't a long term commitment.

> may i suggest, as you have promised that you will code the stuff, to start 
> the coding work as soon as possible. i don't like to make out a testing list 
> but finally see nothing to test. thanks! 

Could you please at least describe, without waiting for anything else, the 
worst challenging scenario you have in mind?
This is a way to make progress.

Thanks,
RD


> 
> - maoke 
>  
>  
>  
> From: Maoke [mailto:[email protected]] 
> Sent: Monday, April 09, 2012 5:52 PM
> 
> 
> To: Sheng Jiang
> Cc: Lee, Yiu; Liubing (Leo); Simon Perreault; [email protected]
> Subject: Re: [Softwires] Path to move forward with 4rdŠ 4rd-U as transparent 
> as MAP-E
>  
>  
> 2012/4/9 Sheng Jiang <[email protected]>
> 
> Without any objection to MAP, we just start to coding 4rd-u and help to 
> improve it.
> 
>  
> fine. we'd love to see it out with good amount of data and good amount of 
> feedback from testing operators. if you invite me to test, i'd love to 
> provide challenging scenarios. ;-) but it doesn't mean we should postpone 
> current process of standardization of MAP, because the result of 4rd-u is 
> still unsure. - maoke 
> 
>  
>  
> Multiple transition mechanisms would co-exist in the world. Neither MAP nor 
> 4rd-U can replace 6rd and DS-Lite. There is no one-suit-all solution. 
> 
>  
> From: Maoke [mailto:[email protected]] 
> Sent: Monday, April 09, 2012 5:39 PM
> 
> 
> To: Sheng Jiang
> Cc: Lee, Yiu; Liubing (Leo); Simon Perreault; [email protected]
> Subject: Re: [Softwires] Path to move forward with 4rdŠ 4rd-U as transparent 
> as MAP-E
> 
>  
>  
> 2012/4/9 Sheng Jiang <[email protected]>
> 
> Please allow me to clarify myself: I support both MAP and 4rd-u.
> 
>  
> MAP has been a good effort with many experts together in IPv4/IPv6 transition 
> area and has been proved by running code. It also need to improve itself 
> before being published as a RFC later. This process will take several months.
> 
>  
> Personally, I think 4rd-u is a little bit behind MAP in maturity aspect for 
> now. But, it is only matter of time. The authors have agreed to remove the 
> architectural conflict, if there was any. We have offered to help on the 
> improvement by implementation.
> 
>  
> Giving time, both solution can be mature enough for operations. IETF does not 
> have to choose one by now. And, these two solutions are not necessary to be a 
> same pace.
> 
>  
> what if, giving time, people can be mature enough to identify 4rd-U is really 
> a flawed donkorse? ;-) if the 4rd-u designers are so confident, why not do 
> the real coding work first and then propose it as a replacement of MAP suite? 
> i don't see any unfairness here. - maoke
> 
>  
>  
> Best regards,
> 
>  
> Sheng
> 
>  
> From: Maoke [mailto:[email protected]] 
> Sent: Monday, April 09, 2012 3:24 PM
> To: Sheng Jiang
> Cc: Lee, Yiu; Liubing (Leo); Simon Perreault; [email protected]
> 
> 
> Subject: Re: [Softwires] Path to move forward with 4rdŠ 4rd-U as transparent 
> as MAP-E
> 
>  
> hi Yiu and others, 
> 
>  
> 1. "both published" does surely conflict with the design goal of 4rd-U 
> itself: 
> 
>    unification of multiple standards. politically speaking, the 4rd-u authors'
> 
>    positition is now quite confusing. 
> 
>  
> 2. operators surely need to choose IPv6 transition mechanisms but such a 
> choice 
> 
>    is based on their demand and understanding to the existing and 
> well-tested, 
> 
>    well-practiced building blocks. i don't think any operators are waiting 
> for 
> 
>    an option of choice they never see and never understand. and the "enough 
> 
>    technical supports" is now still "a check without signature". i do suggest 
> 
>    the 4rd-u authors and supporters say that after making something real. 
> 
>  
>    on the other hand, 4rd-u makes a tight coupling between the address format 
> 
>    and its own header mapping mechanism. this makes an operator unable to 
> have 
> 
>    different choices of transition mechanism as long as it chooses 4rd-u. if 
> 
>    you concern the choice of IPv6 transition mechanism, i do recommend MAP, 
> as 
> 
>    either encapsulation or translation is operatable with MAP address/port 
> 
>    mapping without difficulty. 
> 
>  
> 3. surely MAP still has room of improvement but i don't think there are 
> 
>    architectural uncertainty as 4rd-U has. request on more testing data is 
> 
>    surely appreciated but please think it over again if it is a correct way 
> 
>    to require further data of a counterpart in order to defense a solution 
> 
>    without any data. 
> 
>  
> 4. even there were no MAP at all, current level of 4rd-U is far premature to 
> be 
> 
>    talked about any working group acception.
> 
>  
> the authors and supporters will be appreciated if they talk about the 
> acception 
> 
> after their coding and testing and refinement on the document according to 
> their 
> 
> practical experiences, and understanding the problem again and again. 
> 
>  
> thanks, 
> 
> maoke
> 
>  
> 2012/4/9 Sheng Jiang <[email protected]>
> 
> > I also followed the discussion. I appreciate both teams bought up the
> > technical details for both designs. To be honest, I fail to see which
> > one is better than other (yet). I like the fact that 4rd-u can do what
> > MAP-T does w/o introducing any encap overhead. But I understand the
> > concerns others brought up in the ML such as V-octet. For MAP-T, I like
> > it not introducing any new requirement to the IPv6 header, but I need
> > more published data to determine whether the double-translation would
> > cause any transparent issue or not. For example: I would very
> > appreciate if somebody who trial MAP can publish data such as what
> > applications they have tested, what pass and what fail.
> >
> > This piece of work is very important because once we select it, this
> > technology will stay for long time. I do not want to rush to make a
> > decision and come back to regret in future.
> 
> We would like to see both solution published. So that, operators can choose 
> according to their own networks and preference. There are enormous number of 
> operators, who do not come to IETF meeting or do not participate in ML 
> either. They also need to choose IPv6 transition mechanisms. They should be 
> provided enough technical supports. That's the response of IETF, in our 
> understanding.
> 
> 
> > On 4/6/12 6:00 AM, "Liubing (Leo)" <[email protected]> wrote:
> >
> > >+1
> > >
> > >I've read the endless discussion, and found that is seems the MAP also
> > >has not fully convinced the ISP operation guys.
> > >Since there's explicit controversy, why not just publish them both as
> > >experimental. why we must chose one as a "standard track"? Being a
> > >standard track can eliminate operators' misgiving?
> > >
> > >If the technology is really appropriate, I don't think the
> > "experimental"
> > >status would be a issue for the operators to adopt. Just let the
> > >operator decide by themselves.
> > >
> > >I remembered in the meeting venue, Joel Halpern who chairs Karp and
> > >lisp WG, suggested we publish both of them as experimental, and that
> > >was exactly how they handle the similar situation in their WG. I think
> > >his suggestion is reasonable to be considered. (To Joel: Pardon me if
> > I
> > >improperly quoted your comment, or I just misunderstood your
> > suggestion
> > >).
> > >
> > >> -----Original Message-----
> > >> From: [email protected] [mailto:[email protected]]
> > >>On  Behalf Of Simon Perreault
> > >> Sent: Tuesday, April 03, 2012 8:16 PM
> > >> To: [email protected]
> > >> Subject: Re: [Softwires] Path to move forward with 4rdŠ 4rd-U as
> > >>transparent  as MAP-E
> > >>
> > >> On 2012-04-03 05:40, Ole Trøan wrote:
> > >> > 1) MAP-E supports independence of IPv4 and IPv6 addressing. by
> > >> > using
> > >>hub
> > >> and spoke mode with a separate
> > >> >     mapping rule per subscriber. in this mode e.g only ports could
> > >> > be
> > >> embedded in the address. this is not
> > >> >     possible in a translation solution like 4rd-U.
> > >>
> > >> This mode of operation was suggested last week during the SD-NAT
> > >> presentation. There seemed to be consensus among operators that it
> > >> was not practically feasible.
> > >>
> > >> > 2) a node inside the network will treat translated packets
> > >> > different
> > >>from
> > >> encapsulated ones.
> > >> >     e.g. for the purposes of counting or for applying features.
> > for
> > >> > an
> > >> encapsulated packet both the complete
> > >> >     IPv4 datagram and the IPv6 header is available, and different
> > >>features
> > >> can be applied to both.
> > >>
> > >> In practice, the contents of encapsulated packets are rarely
> > >> inspected by firewalls and other such devices.
> > >>
> > >>
> > >> IMHO at this point we should let the market decide. Publish both as
> > >> experimental. Let's work on something else now.
> > >>
> > >> Simon
> > >> --
> > >> DTN made easy, lean, and smart --> http://postellation.viagenie.ca
> > >> NAT64/DNS64 open-source        --> http://ecdysis.viagenie.ca
> > >> STUN/TURN server               --> http://numb.viagenie.ca
> > >> _______________________________________________
> > >> Softwires mailing list
> > >> [email protected]
> > >> https://www.ietf.org/mailman/listinfo/softwires
> > >_______________________________________________
> > >Softwires mailing list
> > >[email protected]
> > >https://www.ietf.org/mailman/listinfo/softwires
> _______________________________________________
> Softwires mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/softwires
> 
>  
>  
>  
> 
> _______________________________________________
> Softwires mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/softwires

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

Reply via email to