Re: [Softwires] Proposed Unified Address Mapping for encapsulation and double-translation

[Ole Troan <[e-mail]>]

> > i couldn't understand why we "don't see such limitation in 4v6 translation". 4via6 translate the private ipv4 address with 4rd mapping rule while the public ipv4 address with RFC6052 algorithm, right? can the both case work fine if an ISP can only get a fairly long IPv6 prefix, like /40 or even longer?
> >
> > Yes, if ISP has /32 IPv6 address, then how to embed 32 bits IPv4 address and PSID in the first 64 bits?
> 
> only part of the IPv4 address would be embedded into the first 64 bits (not that I think that boundary should be "hard").
> e.g. if the SP has the prefix 10/8. the user the address 10.1.2.3. then only "1.2.3" would be embedded in the address.
> just like what's done for 6rd.
>  
> Thanks, Ole. That’s the point! In this case , you can only have part of the IPv4 address embedded in the first 64 bits. This embedding does work for encapsulation or tunneling, but can NOT work for stateless double translation, which requires the whole IPv4 address+PSID embeded for any IPv6 prefix length.

right, for the outside domain traffic.
putting the state into the IPv6 destination address doesn't affect routing.
so this works fine with the case where you give a user a /56, where you have to have efficient encoding. for the external to the domain traffic you need to encode the full IPv4 address.

cheers,
Ole