Not disagreeing with you, but I would like to point out that the perceived 
security benefits of NAT, which seem to folks like Mr. Engel and Mr. Marquis to 
apply to TCP, apply to a much lesser degree to SCTP. Due to the inherent 
multihoming nature of SCTP, the "default fail closed" behavior of NAT both 
protects less and interferes more than with TCP.

--jhw (sent from my phone)

On Nov 2, 2010, at 8:22, Rémi Després <[email protected]> wrote:

> Hi, James,
> 
> For clarification, is your comment just a positive remark about SCTP, or is 
> it suggesting that SCTP could work without hosts knowing their global 
> addresses.
> (If it is the former, it doesn't go against what I wrote; if it is the 
> latter, I don't see how it can work.)
> 
> Regards,
> RD
> 
> Le 2 nov. 2010 à 16:05, james woodyatt a écrit :
>>> On Nov 2, 2010, at 01:25 , Rémi Després wrote:
>>> ...
>>> SCTP depends on hosts knowing their global addresses, and the same holds 
>>> for SHIM6.
>>> Both are therefore incompatible with all variants of NAT66 as specified 
>>> today.
>> 
>> Actually, SCTP uses IP addresses in pretty much the same way as TCP and 
>> other connection-oriented transport protocols.  From the perspective of a 
>> NAT, however, the requirements to maintain state for SCTP are quite a bit 
>> simpler than for TCP and other protocols. You only need to hold onto the 
>> interior and exterior IP addresses of the association endpoints, unified by 
>> the verification tag for each association.  No port translation is 
>> necessary-- it's not even helpful for the purposes of address amplification. 
>>  The addresses are amplified in the 32-bit verification tag, not the port 
>> numbers.
> 
> 
_______________________________________________
nat66 mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/nat66

Reply via email to