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

["Qi Sun" <[e-mail]>]

Hi Testsuya,


1. If a bunch of entries forming a binding table is not "state", please clarify what is "state" ? And what is so called stateless?

2. You said the entries are "stored", do you mean once it is created it will be never changed? 
If the number of users keeps growing so that it exceeds the number of entries "stored", does BR add new entries? 
If users leave the network, should BR delete those unused entries for saving address resources ?

3. When there are packets inbound on Border Relay(BR), the BR has to look through the binding table for proper destination addresses. That is exactly what lw4over6 does.

Thank you!




Qi Sun

From: Tetsuya Murakami
Date: 2012-06-25 17:36
To: Tetsuya Murakami
CC: sunqi.csnet.thu; Satoru Matsushima; softwires WG; Yong Cui
Subject: Re: [Softwires] [Softwire] draft-ietf-softwire-map-00 does NOT reflect the consensus from the WG
Hi Qi, 


In addition, our MAP implementation can handle such a bunch of entries. Of course, it requires enough memory in order to store all of them in its own memory. But I could not see any other problems except required memory size. Also, each entry is stored statically and so MAP implementation does not maintain the "state" for each entry.
[Qi Sun] If a bunch of entries forming a mapping table is not "state", please clarify what is "state" ? And what is so called stateless?
[Qi Sun] You said the entries are "stored", do you mean once it is created it will be never changed? 



Thanks,
Tetsuya Murakami


On 2012/06/25, at 2:27, Tetsuya Murakami wrote:


Hi Qi, 


Even though there are a bunch of mapping rules, each mapping rule has no state. Since each mapping rules is set manually, BR does not maintain each mapping rule and just store them in its own memory. So, I am not sure why you are calling the "state" if there are a bunch of mapping rules. If you are saying that, do you assume how many entries should be existed in the table in order to call it to the "state". From the implementation perspective, even though there are 10000000 entries in the table, the problem is just how much memory is needed.


Thanks,
Tetsuya Murakami 


On 2012/06/24, at 23:51, Qi Sun wrote:


Hi Satoru,

In MAP 1:1 mode, if there are 10000000 subscribers, there would be 10000000 MAP domains which a BR has to manage. I think that will create a huge mapping table on the BR, which is called 'state' that stateful solutions deal with.

Best Regards!




Qi Sun

From: Satoru Matsushima
Date: 2012-06-25 10:27
To: Lee, Yiu
CC: [e-mail]; Yong Cui
Subject: Re: [Softwires] [Softwire] draft-ietf-softwire-map-00 does NOT reflect the consensus from the WG
Hi Yiu,

No, that's a misunderstanding.
Current MAP specify the case for ea-len is 'zero'. It is 'per-subscriber mapping' in stateless manner, not to introduce 'per-flow NAT binding' or 'per-subscriber state on demand'.

cheers,
--satoru

On 2012/06/25, at 2:32, Lee, Yiu wrote:

> Dear Satoru and MAP-DT
> 
> I echo what Peng and Qiong said. When the WG agreed working on the
> stateless solution, it was very clear stated that the solution would not
> maintain states in the network. If the 1:1 mode changed this, this no
> longer matched the requirements stated in the stateless motivation draft,
> thus, it would disqualify MAP as a solution for the motivation draft.
> 
> AFAIK, the MAP Design Team could propose a change, but such a dramatic
> change by introducing states in the network would require WG approval. I
> would like the chairs to clarify this.
> 
> Thanks,
> Yiu
> 
> 
> On 6/24/12 12:21 PM, "Peng Wu" <[e-mail]> wrote:
> 
>> Hi Qiong, Satoru and all,
>> 
>> I should thank Qiong for pointing this out. I gotta say I'm a bit shocked.
>> If I understand the procedures of IETF correctly, a WG document should
>> reflect the consensus of the WG. MAP is approved by the WG as a
>> stateless solution. As a participator in Softwire, I didn't get the
>> information anywhere that the MAP WG document would cover the
>> so-called 1:1, in fact per-user stateful mode before it was released,
>> not to say discuss in the WG. Don't the WG need to approve such big
>> change anymore?
>> 
>> Now let me provide my impression as an outsider of the MAP DT. You
>> guys make great effort to build the solution, The address composition,
>> the GMA algorithm, the different types of address mapping rules.
>> should be quite difficult to pull together such sophisticated ideas. I
>> guess that's what it takes to achieve the benifits of statelessness.
>> And I admire that, bravo. Then, all of a sudden, you guys are saying,
>> let's apply this sophisticated method to the different problem, by
>> dropping quite some comlexity and twistting the mechanism a bit, seems
>> it may work. Considering the problem are now solved in a more pure and
>> clear way, I'm sorry but I CANNOT follow the logic here.
>> 
>> On Sun, Jun 24, 2012 at 2:00 PM, Satoru Matsushima
>> <[e-mail]> wrote:
>>> Hi Qiong,
>>> 
>>> I'm disagree with your opinion.
>>> 
>>> 1. Recent changes in draft-ietf-softwire-map-00 has been discussed in
>>> the DT.
>>> 2. MAP just covers so called '1:1 mode' with most granular mapping rule
>>> for CEs provisioning, which is as one of its characteristics.
>>> 3. The motivation draft does not restrict that as you stated, just
>>> 'assumed', it's neither 'MUST' nor 'SHOULD'.
>>> 
>>> Best regards,
>>> --satoru
>>> 
>>> 
>>> On 2012/06/24, at 14:35, Qiong wrote:
>>> 
>>>> Hi all,
>>>> 
>>>> As we all know, once an individual draft is adopted as a WG draft, it
>>>> is owned by the whole WG, rather than just the editors. Just as Remi
>>>> said, the normal procedure to follow is to reach WG consensus _before_
>>>> posting a newly edited version.
>>>> 
>>>> From draft-mdt-softwire-mapping-address-and-port-03 to
>>>> draft-ietf-softwire-map-00, there are several changes between them. In
>>>> particular, the newly introduced "1:1 mode", which decouples IPv4 and
>>>> IPv6 addressing, has never been discussed openly in the WG mailing
>>>> list, or even in the MAP design team either.
>>>> 
>>>> Actually, this "1:1 mode" is against the stateless-4v6-motivation
>>>> draft. The motivation draft has clearly defines the "Stateless 4/6
>>>> solution" as follows:
>>>> 
>>>> Stateless 4/6 solution denotes a solution which does not require any
>>>> per-user state (see Section 2.3 of [RFC1958]) to be maintained by any
>>>> IP address sharing function in the Service Provider's network. This
>>>> category of solutions assumes a dependency between an IPv6 prefix and
>>>> IPv4 address.
>>>> 
>>>> AFAIK what the WG has adopted MAP related draft is
>>>> draft-mdt-softwire-mapping-address-and-port-03, NOT
>>>> draft-ietf-softwire-map-00. And the stateless solution should ³response
>>>> to the solution motivation document² according to the Softwire charter.
>>>> That means draft-ietf-softwire-map-00 IS NOT QUALIFIED to be a WG draft.
>>>> 
>>>> We can all recall that our softwire WG has worked on stateless
>>>> solutions for more than one and a half years, and we have achieved a
>>>> lot of work which has been documented in charter, stateless motivation,
>>>> 4rd-varients, MAP-03, etc. AFAIK all the authors have kept the basic
>>>> "stateless" principle and the MAP design team is also working on it
>>>> together to find a better algorithm, address format, etc. So it is
>>>> really not appropriate to make such changes when MAP is adopted as a WG
>>>> item in such a short time.
>>>> 
>>>> From this perspective, draft-ietf-softwire-map-00 can only be regarded
>>>> as draft-XX-softwire-mapping-address-and-port-04. It is not even the
>>>> output of MAP design team.
>>>> 
>>>> Best wishes
>>>> 
>>>> ==============================================
>>>> Qiong Sun
>>>> China Telecom Beijing Research Institude
>>>> 
>>>> 
>>>> Open source code:
>>>> lightweight 4over6: http://sourceforge.net/projects/laft6/
>>>> PCP-natcoord: http://sourceforge.net/projects/pcpportsetdemo/
>>>> ===============================================
>>>> 
>>>> 
>>>> _______________________________________________
>>>> Softwires mailing list
>>>> [e-mail]
>>>> https://www.ietf.org/mailman/listinfo/softwires
>>> 
>>> _______________________________________________
>>> Softwires mailing list
>>> [e-mail]
>>> https://www.ietf.org/mailman/listinfo/softwires
>> _______________________________________________
>> Softwires mailing list
>> [e-mail]
>> https://www.ietf.org/mailman/listinfo/softwires

_______________________________________________
Softwires mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/softwires
_______________________________________________
Softwires mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/softwires