On Thu, Nov 13, 2014 at 2:02 PM, Dino Farinacci <[email protected]> wrote:
> On Thu, Nov 13, 2014 at 12:06 AM, Dino Farinacci <[email protected]> wrote:
>>>> Specifically, one could imagine the NVE copying the IP1 DSCP codepoint 
>>>> into the IP2 header. Alternatively one could
>>>
>>> Exactly. Thanks Benson.
>>
>> Dino, yes, but this requires DPI at the NVE, as we already talked
>
> No it doesn't.
>
>> about in the draft as well as configuration to use diffserv.
>> RFC7348 does not mention any requirement for NVE to look beyond Inner
>> Ethernet Header into the Original Ethernet Payload in order to make
>> its encapsulation.
>
> This is very simple. Let me sequence the steps on the encapsulator as well as 
> the decapsultor:
>
> (1) A 802.1Q frame arrives at an encapsulator.
> (2) The encapsulator will encapsulate the frame with an outer IP header. It 
> copies the EXP bits from the frame to the outer IP header TOS bits.
> (3) The decapsulator receives the IP packet, it strips the outer IP header 
> and DOESNOT copy the TOS bits to the EXP bits of the frame because the TOS 
> field could have been changed in transit. The EXP bits, set by the 
> originating source are perserved.
>

Sorry there are no EXP bits mentioned in RFC 7348. MPLS is out of scope.
EXP is 3 bits long, DSCP is 6 bits and dividing it into two 3 bit
pieces, I am not sure if David will like it.

See also Brian Carpenter's comments.

Regards,

Behcet

> Dino
>
>>
>> Regards,
>>
>> Behcet
>>
>>>
>>> Dino
>

_______________________________________________
nvo3 mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/nvo3

Reply via email to