Re: Issue: Support for IPv4 - IPv6 movements (#13)
Stephen Kent <[email protected]>
| Newsgroups | gmane.ietf.mobike |
|---|---|
| Message-ID | <p06110403bd05dfbd8712@[128.89.89.75]> |
At 11:25 AM +0200 6/26/04, Francis Dupont wrote: > In your previous mail you wrote: > > The SCTP requirement is to map sets of S/D addresses to a single SA > via the SPD, which we do. > >=> this is about inner addresses aka traffic selectors as the SPD >knows only them, isn't it? yes. > The RFC also calls for mapping traffic > from a set of destination addresses to a single SA, via the SAD. This > is something that 2401bis accommodates, in the discussion of inbound > traffic processing in Section 5.2, step 2, so long as we interpret > the phrase "If the packet appears to be IPsec protected and it is > addressed to this device" as allowing multiple addresses for the > IPsec device. We will add a note to clarify this. > >=> fine, so the inbound case is solved by the introduction of >suitable flexibility. But the outbound case is not and IMHO >to require the support of multiple simultaneous outer addresses >is a bad idea. Of course in an IKE context (as MOBIKE) IPsec SAs >come by pairs and we can't ignore the outbound issue... > Yes, outbound is problematic. 2401/bis is not comprehensive in this regard, anyway. Note that we allude to the problem of finding a security gateway that services a set of hosts, but that we don't say exactly how to solve the problem, since there is not Wg agreement on how to do this. Steve