[Netext] Keep missing the point .. Re: next steps for netext

Hi George,

I agree with your conclusion and I believe that the important thing for 
the IETF community to answer is how far we should take the PMIPv6 protocol.

The question asked at the end of the Netext BOF was whether people 
believe that CMIP or PMIP should be used to solve the multi-homing 
problem (the final question asked by Jari). There was no rough consensus 
about this in either direction.

As I see it there are two options regarding multi-homing and flow 
mobility support for PMIPv6: Either PMIPv6 is transparent for the host 
and everything is handled by CMIP and associated protocols (this is 
something that has been standardized in the mext WG), or necessary 
additions are made to the PMIPv6 protocol and the host to support it. We 
wrote an Internet draft 
(http://tools.ietf.org/html/draft-larsson-netext-pmipv6-sma-flow-mobility-00) 
were we tried to give an idea of how such a solution could look like for 
PMIPv6. This requires modifications of the host in the sense that 
distributions of outgoing flows over different network interfaces must 
be handled by the host.

Regards
Conny Larsson

George Tsirtsis wrote:
> Folks,
>
> What people keep missing is that 3GPP, 3GPP2, WiMax etc, are all
> PROPRIETARY systems as far as the IETF is concerned. The IETF is
> tasked with making sure that the Internet as a whole holds together.
> The IETF should only standardise protocols/tools requested by SDOs if
> they do not cause harm to the general Internet. This is paramount for
> the long term sustainability of the Internet.
>
> Independently from "architectural" viewpoints, it is perfectly
> reasonable for a given SDO to decide to implement a given technology
> (e.g., PMIP) in their CLOSED System. PMIP for example can work in the
> context of 3GPP (as GTP did for many years) since they define a
> vertical system, from PHY and LINK layers, to Network and even
> Application layers. In that case, they are free to, for example,
> implement whatever Link Layer triggers are necessary to make PMIP work
> in their system.
>
> It is another thing altogether, however, for the IETF to standardise
> such a technology, when here we do not have control over link layers.
> On the contrary what the IETF and the Internet are based on is an
> explicit assumption that "any" link layer should work with minimum
> requirements from the IP layer, all of which are described in host
> requirements
>
> What PMIP proponents have been asking the IETF to do is define a
> technology that will break an RFC4294 host.
> This is simply not acceptable.
>
> George
>
> On Tue, Apr 7, 2009 at 11:55 AM, Domagoj Premec
> <domagoj.premec.ext at nsn.com> wrote:
>   
>> Hi Hesham, all,
>>
>>     
>>> What's being said here is "PMIP
>>> doesn't support advanced cases therefore we need something
>>> new", which is completely bogus because PMIP was not designed
>>> to support those cases because we already have something else
>>> for supporting those cases. Hence my reference to 4830...
>>>
>>>       
>> Both 3GPP and WiMAX adopted PMIP6 as means for inter-access handover. Such
>> large scale managed networks are the main consumers of PMIP6 so we shouldn't
>> be ignoring what they're doing. Saying to use MIP for inter-access
>> handovers/multihoming is not helping as they have really set their mind on
>> using PMIP6 and they're already doing this. Without going into the reasoning
>> behind such a decision, the fact is that PMIP6, as defined in RFC 5213, has
>> issues with inter-access handovers. I think it is better to fix those PMIP6
>> issues in the IETF then to let each SDO come up with their own way of
>> dealing with this.
>>
>> domagoj
>>
>>
>>     
>>> -----Original Message-----
>>> From: netext-bounces at mail.mobileip.jp
>>> [mailto:netext-bounces at mail.mobileip.jp] On Behalf Of ext
>>> Hesham Soliman
>>> Sent: 07. travanj 2009 11:14
>>> To: Marco Liebsch
>>> Cc: netext at mail.mobileip.jp; Basavaraj.Patil at nokia.com
>>> Subject: Re: [Netext] next steps for netext
>>>
>>>
>>>       
>>>>> => We're discussing a problem statement for the BoF, so
>>>>>           
>>> 4830 is quite
>>>       
>>>>> appropriate as a reference for why people wanted PMIP in
>>>>>           
>>> the first place.
>>>       
>>>> Yes, a problem statement about an existing solution for
>>>>         
>>> network-based
>>>       
>>>> mobility management (PMIPv6) to support advanced use cases.
>>>>         
>>> => Where is that PS?
>>>
>>> ?  Not the problem of
>>>       
>>>> existing solutions for localized mobility, as RFC4830 does. I don't
>>>> think NetExt aims at turning PMIPv6 into a host-based mobility
>>>> protocol again.
>>>>         
>>> => I don't know what the goal is based on the BoF discussion
>>> and this discussion on the list. The reason I don't know that
>>> is that no one is
>>> saying:
>>> Here is the state of the art (including BOTH MIP and PMIP)
>>> and here is why they don't work, therefore we need to do
>>> something new.
>>> That's what a PS should say. What's being said here is "PMIP
>>> doesn't support advanced cases therefore we need something
>>> new", which is completely bogus because PMIP was not designed
>>> to support those cases because we already have something else
>>> for supporting those cases. Hence my reference to 4830...
>>>
>>> I'm going to retire from the discussion till someone makes
>>> the argument above. Because I feel that if there is no such
>>> argument then the discussion is not a good use of time.
>>>
>>>
>>>       
>>>>> => Yes, additional software like the existing MIP. I don't
>>>>>           
>>> understand
>>>       
>>>>> the motivation for creating something new. I've detailed
>>>>>           
>>> this in my
>>>       
>>>>> previous emails and the points remain unanswered.
>>>>>
>>>>>           
>>>> Nobody talks about adding a piece of client mobility management
>>>> software, such as a MIP client, to the mobile. That seems to be a
>>>> wrong interpretation of what NetExt wants to do.
>>>>         
>>> => You're missing the point. Why isn't anybody talking about
>>> MIP??? That's what a PS should say.
>>>
>>> Hesham
>>>
>>>       
>>>> marco
>>>>
>>>>
>>>>         
>>>>> Hesham
>>>>>
>>>>>
>>>>> and obviously also according to RFC4831. We should
>>>>>
>>>>>           
>>>>>> distinguish modification for mobility management, which
>>>>>>             
>>> is not what
>>>       
>>>>>> we want to do, and enabling local functions, such as the use of
>>>>>> multiple interfaces, where you have to add routes, configure
>>>>>> interfaces etc. to enable the use case.
>>>>>>
>>>>>> marco
>>>>>>
>>>>>>
>>>>>> Hesham Soliman schrieb:
>>>>>>
>>>>>>             
>>>>>>>               
>>>>>>>>> => I think you're distinguishing between modified and
>>>>>>>>>                   
>>> unmodified
>>>       
>>>>>>>>> based on whether the modification affects IETF
>>>>>>>>>                   
>>> protocols or not.
>>>       
>>>>>>>>> That's fine, but for those that did not attend the BoF, our
>>>>>>>>> conclusions were that you will either end up modifying
>>>>>>>>>                   
>>> the host
>>>       
>>>>>>>>> (i.e. Adding new SW) and adding new signalling on
>>>>>>>>> L2 to handle the multihoming cases, or you can stick with MIP.
>>>>>>>>> My opinion, and at least half the room's was that this
>>>>>>>>>                   
>>> is not a
>>>       
>>>>>>>>> good way to
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>                   
>>>>>>>> We can discuss what implies host changes and why this is such a
>>>>>>>> big deal anyway. But I don't think that half the room was in
>>>>>>>> agreement as such. At least from the count of hands
>>>>>>>>                 
>>> raised for the
>>>       
>>>>>>>> question about whether we should work on multihoming
>>>>>>>>                 
>>> and intertech
>>>       
>>>>>>>> handovers it was about 18-9 (in favor).
>>>>>>>>
>>>>>>>>
>>>>>>>>                 
>>>>>>> => I think the question changed a few times :) The final
>>>>>>>               
>>> question
>>>       
>>>>>>> (which was specific to whether people wanted to work
>>>>>>>               
>>> with this or
>>>       
>>>>>>> not) showed a split room.
>>>>>>> The host changes are not a big deal but they are mentioned here
>>>>>>> because they were one of the very few reasons for using
>>>>>>>               
>>> NETLMM in
>>>       
>>>>>>> the first place: Host changes (adding new SW to the host) and
>>>>>>> signalling from the MN. You can see them in section 4 of
>>>>>>>               
>>> RFC 4830.
>>>       
>>>>>>>
>>>>>>>               
>>>>>>>>> go and that it's better to stick with MIP instead of
>>>>>>>>>                   
>>> relying on a
>>>       
>>>>>>>>> solution that will:
>>>>>>>>> - work on some L2s and not others,
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>                   
>>>>>>>> We have no ability to specify changes to L2 anyway. However if
>>>>>>>> some L2s do have such capability, why would we not specify how
>>>>>>>> multihoming would work in such scenarios.
>>>>>>>>
>>>>>>>>
>>>>>>>>                 
>>>>>>> => Because we already have a solution that works in all L2s,
>>>>>>> doesn't require
>>>>>>> L2 changes that we don't control and doesn't cause
>>>>>>>               
>>> confusing host
>>>       
>>>>>>> config solutions.
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>               
>>>>>>>>> - Require additional SW and config on the host which
>>>>>>>>>                   
>>> is not done
>>>       
>>>>>>>>> anywhere today, and
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>                   
>>>>>>>> Why is there as much of a concern about additional SW
>>>>>>>>                 
>>> or config?
>>>       
>>>>>>>> After all everything (or most of it anyway) that we do
>>>>>>>>                 
>>> in the IETF
>>>       
>>>>>>>> requires configurations, protocol changes, changes to SW etc.
>>>>>>>> There is nothing unique about this.
>>>>>>>>
>>>>>>>>
>>>>>>>>                 
>>>>>>> => Sure, but see above and see RFC 4830. These were the
>>>>>>>               
>>> reasons for
>>>       
>>>>>>> using PMIP in the first place. We can't have it both
>>>>>>>               
>>> ways, or can
>>>       
>>>>>>> we? :)
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>               
>>>>>>>>> - Require L2 signalling which is out of our scope in IETF
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>                   
>>>>>>>> Agree. We cannot do anything about L2s. But we could specify
>>>>>>>> (informatively)
>>>>>>>> what L2 capabilities would enable these features if that is the
>>>>>>>> conclusion that we arrive at.
>>>>>>>>
>>>>>>>>
>>>>>>>>                 
>>>>>>> => We can, but I don't know why we should do that when a
>>>>>>>               
>>> solution
>>>       
>>>>>>> already exists. Especially when the alternative is inferior in
>>>>>>> terms of its applicability.
>>>>>>>
>>>>>>> Hesham
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>               
>>>>>>>> -Raj
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>                 
>>>>>>>>> This is why I think we should stick with the charter
>>>>>>>>>                   
>>> that Jari sent.
>>>       
>>>>>>>>> Hesham
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>                   
>>>>>>>>>> I share the same view with respect to all your other
>>>>>>>>>>                     
>>> comments.
>>>       
>>>>>>>>>> We need to seperate the cases of flow mobility vs
>>>>>>>>>>                     
>>> basic handoff
>>>       
>>>>>>>>>> as allowed in 5213. In the BOF, we discussed mostly flow
>>>>>>>>>> mobility and not the basic handoff cases supported
>>>>>>>>>>                     
>>> today in 5213.
>>>       
>>>>>>>>>> Sri
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> _______________________________________________
>>>>>>>>>> NetExt mailing list
>>>>>>>>>> NetExt at mail.mobileip.jp
>>>>>>>>>> http://www.mobileip.jp/mailman/listinfo/netext
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>                     
>>>>>>>>>                   
>>>>>>> _______________________________________________
>>>>>>> NetExt mailing list
>>>>>>> NetExt at mail.mobileip.jp
>>>>>>> http://www.mobileip.jp/mailman/listinfo/netext
>>>>>>>
>>>>>>>
>>>>>>>               
>>>>>
>>>>>           
>>>>         
>>> _______________________________________________
>>> NetExt mailing list
>>> NetExt at mail.mobileip.jp
>>> http://www.mobileip.jp/mailman/listinfo/netext
>>>
>>>       
>> _______________________________________________
>> NetExt mailing list
>> NetExt at mail.mobileip.jp
>> http://www.mobileip.jp/mailman/listinfo/netext
>>
>>     
>
> _______________________________________________
> NetExt mailing list
> NetExt at mail.mobileip.jp
> http://www.mobileip.jp/mailman/listinfo/netext
>   
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.mobileip.jp/pipermail/netext/attachments/20090407/bc14e033/attachment-0001.html>