Re: [addr-select-dt] slide to 6man presentation

[Arifumi Matsumoto <[e-mail]>]

On 2010/07/26, at 23:48, [e-mail] wrote:

> Hi,
> 
>>> The synthesized addresses probably should be prioritized below public
>> IPv4. When comparing synthetic IPv6 to private IPv4, its maybe a matter
>> of taste - assuming ALGs are in place in NAT44 and NAT64 (general
>> preference for IPv6 could win also).
>> 
>> So, you mean
>> global IPv4 should be prioritized than synthesized IPv6 address.
>> but, private IPv4 may not be prioritized than synthesized IPv6
>> address.
> 
> Yes, truly global IPv4 should be higher (hence avoids all NATtting).
> 
>> The RFC3484 revision draft also include a proposal that private IPv4
>> address should be scoped global. So, if we follow it, we should not
>> differentiate the preferences of global and private.
> 
> Hence private IPv4 would be considered higher than synthesized IPv6 using WKP? That is fine, probably NAT44 is more reliable than NAT64 (and may have better ALG support).
> 
>>> The prefix used for synthetized IPV6 addresses (if not using well-
>> known-prefix (WKP)) would need to be dynamically detected..
>> 
>> The host itself does not need to detect it, if the NSP distribute an
>> appropriate policy table for it.
>> Re. WKP, we should have to put it in the default policy table.
> 
> Ok WKP could be low, and NSP could be distributed with policy table. And isn't it so that if some module on a host learns NSP (e.g. DNS resolver or DHCP or RA), it can modify the policy table accordingly (hence locally generated policy).


Of course, I do not preclude such an approach.
But I think, it is better if such a configuration is enforced from the network.
Just like the privacy address preferences should be configurable per network.