Re: Comments on draft-koodli-ipv6-in-mobile-networks-02.txt

[Behcet Sarikaya <behcetsarikaya@yahoo.com>]

Hi Rajeev,




> Hi Rajeev,
>   My comments on 
> Section 3.2 NAT 
> Placement in the mobile networks
> 
> This section nicely narrates 
> the story of GI-DS-Lite as the centralized
> NAT versus DS-Lite as the 
> distributed DS-Lite. It reads nicely but I found in
> it too much emphasis 
> on billing and accounting aspects while it lacks emphasis
> on other 
> aspects (more meaty? aspects at least as much as I am 
> interested).

> DS-Lite is not presented as a distributed NAT solution in 
> the document.


I thought it should be.


>Many mobile providers do care for subscriber management. 
> For them, it is
important to be able to do harmonized billing and accounting. 
> With respect
to NAT itself, which introduces binding state between addresses, 
> it is
important to be able to harmonize NAT state with subscriber billing 
> and
accounting.

This is OK as I said before. But it is one kind of consideration when deploying IPv6 in mobile networks. I was asking about some other considerations you have just ignored.

> 
> For example, one very important feature 
> of DS-Lite is to have an IPv6-only
> network thus savings on IPv4 
> addresses. This is not mentioned.
> Next, why currently mobile operators 
> do not have much interest on DS-Lite is
> not 
> mentioned.

>Clarification: this ID is *not* about evaluating pros and cons 
> of each of
the available transition mechanism. It is about considerations 
> that arise
>when deploying IPv6 in mobile networks.

Please see my point above. You mean savings in IPv4 address usage is not a consideration?


>To your point about 
> DS-lite itself, it is perfectly fine in residential
broadband networks. For 
> mobile networks, having a tunnel termination at the
UE (host) is considered 
> undesirable.


Eaxctly. I am asking these types of considerations to be stated in the distributed NAT text.


> The 3GPP TR on IPv6 transition looks at
some of the variants 
> FYI, and that document may be a good place for your
questions. 
> 

Thanks, I know this document.


> 
> On the centralized NAT or GI-DS-Lite, firstly, the 
> identification of
> GI-DS-Lite as the centralized NAT approach is very 
> useful because it was not
> clear reading the GI-DS-Lite draft.
> 
> 
> However, one important aspect of GI-DS-Lite is that it no longer 
> requires an
> IPv6-only network and therefore no IPv4 address savings 
> which is not mentioned
> in the draft.

I did not follow this. Care 
> to elaborate?


GI DS Lite unlike DS Lite is not based on IPv6 only network premise, that is an important consideration when deploying IPv6 in mobile network. I was simplying asking these things to be stated in the centralized NAT discussion.
Also the fact that centralized NAT makes MN to MN communication to take longer paths.



Regards,

Behcet