Re: Issue: Support for IPv4 - IPv6 movements (#13)

  


> In your previous mail you wrote:
> 
>    Yes, it is about the outer addresses.
>    
>    Originally when I sent the e-mail I was thinking about movement
>    from an IPv4 address to an IPv6 address. But of course there's
>    a similar issue with multihoming, as discussed in RFC 3554.
>    
>    As you Francis and Mohan pointed out, supporting multiple
>    outer and inner addresses is already in current specifications.
> 
> => but there is a real difference:
>  - it is expected that multiple inner addresses will be supported
>    by nearly every implementations (the only way to avoid the support
>    is to require only single traffic selector pairs but this is like
>    to reject not initial exchanges: legal but only for dumb
>    implementations).
>  - multiple outer addresses for IPsec SAs are very complex to
>    implement. BTW, does someone know a complete implementation
>    of RFC 3554?
> 
>    My expectation is that this should also work with MOBIKE, with
>    both movements and multihoming. I propose that we close this
>    issue until we find some evidence that we have a problem to
>    support this :-)
>    
> => I believe that there is a real problem but the problem is partially
> outside the scope of MOBIKE. I propose to *not* require the support
> of multiple simultaneous outer addresses for IPsec SA without a strong
> support from the IPsec WG (as something specific in the 2401bis for
> instance).
> 
I miss something here. I thought RFC 3554 has the support of the IPsec WG.
All we are saying that this address set can change. And changing addresses
is the core of  MOBIKE.

-mohan

> Thanks
> 
> [email protected]