On 4/28/2015 10:36 AM, Templin, Fred L wrote:
> Hi Joe,
> 
>> -----Original Message-----
>> From: Joe Touch [mailto:[email protected]]
>> Sent: Tuesday, April 28, 2015 10:29 AM
>> To: Templin, Fred L; Tom Herbert
>> Cc: [email protected]
>> Subject: Re: [Int-area] Why combine IP-in-UDP with GUE?
>>
>>
>>
>> On 4/28/2015 10:13 AM, Templin, Fred L wrote:
>>> Hi Tom,
>>
>> ...
>>>>>> Without any optional fields or flags, the difference with GUE is an
>>>>>> additional four byte header between the UDP header and the
>>>>>> encapsulated IP header. For IPv4 that header is 0x00040000, and for
>>>>>> IPv6 the header is 0x00290000
>>
>> ...
>>> This loops back again to whether the first four bits of the payload could
>>> be used such as:
>>>
>>>   4 - a native IPv4 packet
>>>   6 - a native IPv6 packet
>>>   X - a GUE-encapsulated packet
>>
>> It loops back to what the service is.
>>
>> Encapsulation of IP is a service.
>>
>> There is no justification for differentiating between encapsulation of
>> IPv4 and IPv6 at the GUE or IP-in-UDP layer.
>>
>> Nothing on the path should need to see what kind of IP packet it is. If
>> it does, it can easily parse the first for bits of the IP-in-UDP or GUE
>> payload - NOT the UDP payload.
> 
> What I mean to say is "the first four bits immediately following the UDP
> header". At least, that is what IP-in-UDP and AERO say. Were you thinking
> there would be some other four bits?

It's more clear if we just say "the first four bits of the IP header" -
wherever the encapsulation puts that header.

If that's the case, there's no issue in interpreting X=4 or X=6. We
can't assign meanings to other values without a standards-track doc.

We CAN put a flag bit/field in a shim header that explains that the
service carries more than just IP, and the flag bit/field indicates
"next protocol=IP" or "next protocol=signal", etc.

Joe


_______________________________________________
Int-area mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/int-area

Reply via email to