Re: [Softwires] [Softwire] draft-ietf-softwire-map-00 does NOT reflect the consensus from the WG

[Satoru Matsushima <[e-mail]>]

On 2012/06/27, at 16:43, Peng Wu wrote:

> On Wed, Jun 27, 2012 at 2:57 PM, Satoru Matsushima
> <[e-mail]> wrote:
>> 
>> On 2012/06/27, at 15:38, Peng Wu wrote:
>> 
>>>> Oh, you don't argue that OSPF covers an use case which is also covered by RIP. So then why are you arguing that an use case of MAP is eventually same with the LW46 use case?
>>> I'm clearly saying they have different use cases, but that's not the
>>> point. Let me repeat. If I want RIP, you cannot just place RIP into
>>> OSPF,
>> 
>> Agree on that it's not what I'm intended to. MAP thus never put DHCPv4 over IPv6, nor PCP into its specification. Please keep your mind in peace.
>> 
>>> put an OSPF "face" on it, and force me to use the OSPF "suite"
>>> while the essence of the protocol I'm using is still RIP.
>> 
>> Not to force, MAP uses its MAP protocol to an use case which also could be covered by LW46's DHCPv4 over IPv6, or PCP. Correct?
>> 
> Yes, there are mutliple choices for the provisioning protocol. But the
> essence here is 1.no v4-v6 address coupling, and thereby 2.explicitly
> provisoin the v4 address and port set. I would say the ORIGINAL MAP
> fits with neither points here.
> 
> BTW,  if I may, my suggestion on MAP to deal with the situation of EA
> bit=0 is, just say in this case there is no algorithmic address
> mapping so it's not consistent with general case or the original
> motivation, and thereby not covered. I believe it is not the main
> scenario you want to cover and this way you keep MAP clean.

Thank you for your technical suggestion. 
But it couldn't be an option because MAP node finds its BMR in longest match basis between end-user ipv6 prefix and the rule-ipv6-prefix whether the ea-len is zero or not in the BMR.

FWIW, my suggestion is that LW46 adopts MAP provisioned CE to be a kind of TI which connected to a TC.

cheers,
--satoru