Re: [MBONED] APPSDIR review of draft-ietf-mboned-64-multicast-address-format-01

[Stig Venaas <[e-mail]>]

On 5/14/2012 1:50 PM, Lee, Yiu wrote:
> Hi Brian,
>
> Sorry for getting back late. I read Bert's answers to your questions. His
> answers are inline with my answers. Most information are statically
> configured. For example: Ch1 is statically configured to 224.1.2.3 via OOB
> mechanism. If the STB is IPv4 only, it will only use IPv4 mcast address.
> It won't use the address format defined in this draft. In my mind, the
> most common deployment will use the same IPv6 prefix which will be
> statically configured in the AF.

Right, so that was my main concern with the draft. Is static
configuration like this sufficient, or are there more generic cases
where one needs to know that it is translated from IPv4 without a
priori configuration. The flag bit (or a well-known prefix) is only
needed in the latter case. I know there were some scenarios where
someone found the latter to be advantageous though.

Stig

>
> Regards,
> Yiu
>
>
> On 5/10/12 2:03 PM, "Brian Haberman"<[e-mail]>  wrote:
>
>> Hi Yiu,
>>       Let me ask a few questions...
>>
>> On 5/9/12 10:52 PM, Lee, Yiu wrote:
>>> Hi Carsten,
>>>
>>> Thanks very much for reviewing the document. I just want to add a point
>>> to
>>> your question about how applications decide when to use this multicast
>>> address format. In fact, they don't. Imagine a use case where a legacy
>>> IPv4 IP-TV receiver (an app) wants to join a channel which is
>>> broadcasted
>>> in IPv6. The app will continue to send the igmp-join (say 224.1.2.3).
>>
>> How does the IPv4 IP-TV know to join 224.1.2.3?
>>
>> How is 224.1.2.3 advertised to the IPv4 IP-TV clients if the content is
>> generated by an IPv6 source?  Does the source need to be configured to
>> use one of these IPv4-in-IPv6 multicast addresses?
>>
>>> There will be a function in the network which is statically configured
>>> that when it receives a igmp-join, it would covert to a corresponding
>>> mld-join. The IPv6 address in the join message will follow what is
>>> described in this draft. This Adaptive Function is transparent to the
>>> application and managed by the network.
>>
>> Are you limiting this approach to only mapping at the IGMP/MLD protocols?
>>
>> How does your Adaptive Function know which IPv6 multicast prefix to use
>> when mapping the IPv4 multicast address in the IGMP Report message to MLD?
>>
>> Regards,
>> Brian