Re: Next steps on Extension Header Insertion

[[e-mail]]

Sander,

> 
>> To paraphrase Michael Moore: "If you as a man is against gay marriage, then you do have the option of not marrying another man".
> 
> Everybody is fully entitled to their own way of life and love :)  But the analogy doesn't fit...
> 
>> In this context; you have the choice of not deploying any packet mangling middleboxes.
> 
> If the mangled packets can be contained in one administrative domain then that's fine. I think that this is the point where the disagreement starts. Standards are about interoperability: everything that goes beyond your administrative domain. I don't care what people do to packets on their network, as long as interoperability doesn't suffer: as long as other operators don't even notice.
> 
> So I think the standard should say "this is what externally observable behaviour should/must be" with the understanding that "if you deviate from the standard internally, you must make sure the rest of the world doesn't notice".
> 
> I think that's what interoperability is about: your network, your rules, but between networks we use standards-based IPv6.

What you are saying is something like:

"The IPv6 header fields described as immutable in RFC4302 MUST NOT be changed by the network. If within an administrative domain any of the immutable fields are changed, they MUST be restored on exit from the domain."

Correct?

Cheers,
Ole

PS: As an operator would you be happy if I sent packets with NH=59, SA=:: with an integrity check covering all fields apart from the HLIM field, and everything else beyond the IPv6 header (40 bytes) was encrypted?