Le 18 août 2011 à 10:42, Washam Fan a écrit :

> Hi Remi
> 
> 2011/8/18 Rémi Després <[email protected]>:
>> 
>> Le 18 août 2011 à 09:18, Washam Fan a écrit :
>> 
>>>>> It seems to me, when delegating CE ipv6 prefix, a longest match might
>>>>> be used.
>>>> 
>>>> OK, but, again, if a realistic use case is available where longest match 
>>>> is indeed REQUIRED, there is no problem to impose longest match.
>>>> What is missing so far is this use case.
>>> 
>>> can i cite Tetsuya's examples to elaborate that? assume we have three
>>> below rules
>>> 1.{2408:db8:100::/40, 10.10.1.0/24, 48}
>>> 2.{2408:db8:200::/40, 10.10.2.0/24, 48}
>>> 3.{2408:db8::/32, 10.10.0.0/24, 48}
>> 
>> To be complete, the example should IMHO show which CPE's have which IPv6 
>> prefixes, and with which IPv4 prefixes derived.
>> OK?
> OK.
> 
>    CE prefix                                   derived IPv4
> 1.{2408:db8:100::/40, 10.10.1.0/24, 48}
> 2048:db8:101::/48                          10.10.1.1
> 2.{2408:db8:200::/40, 10.10.2.0/24, 48}
> 2048:db8:201::/48                         10.10.2.1
> 3.{2408:db8::/32, 10.10.0.0/24, 48}
> 2408:db8:301::/48                         10.10.0.3+ports
> 
>    2408:db8:302::/48                         10.10.0.3+ports
> 
> note that rule 3 responds to many CE prefixes and derived ipv4
> addresses (plus ports)
> 
> Is that clearer?

I will look at it, and see what I think.
Thanks.
RD


> Thanks,
> washam
>> Regards,
>> RD
>> 
>>> 
>>> if a 48 CE prefix is delegated, rule 1 and 2 should be checked against
>>> before rule 3. Although in this case, you can not assign 10.10.0.1 and
>>> 10.10.0.2 ipv4 addresses to a CE.
>>> 
>>> Thanks,
>>> washam
>> 
>> 
>> 


_______________________________________________
Softwires mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/softwires

Reply via email to