Re: issue 3: nat traversal
Francis Dupont <[email protected]>
| Newsgroups | gmane.ietf.mobike |
|---|---|
| Message-ID | <[email protected]> |
In your previous mail you wrote: In the meantime, I wonder if we could get any leverage out of a "NAT expected" bit -- carried in a secured part of the protocol. This would need to be administratively configured. It would probably have to default to "on". => in fact I share your opinion with only a small difference: I'd like to get a "NAT forbidden" bit. If both peers have that bit cleared you would do NAT prevention; otherwise, you would do NAT traversal if a NAT is detected. => NAT prevention and NAT detection are the same until there is a NAT in the path. One is needed in mobike because as a side effect the peer addresses are protected (because they are duplicated from the header to the content of IKE message). Note the choice really exists only for the initiator, the responder always knows when there is a NAT in the path (i.e., it is not really symmetrical). Thanks [email protected]