Paul,

Thank you very much for the quick response.
Please see below of the response. If there is still confusion, can we have  a 
zoom chat?

Linda

From: Paul Wouters <[email protected]>
Sent: Friday, August 15, 2025 2:20 PM
To: Linda Dunbar <[email protected]>
Cc: ipsecme-chairs <[email protected]>; IPSEC <[email protected]>; Deb 
Cooley <[email protected]>
Subject: Re: Follow-up on adoption options for 
draft-dunbar-ipsecme-lightweight-authenticatee


On Fri, Aug 15, 2025 at 2:03 PM Linda Dunbar 
<[email protected]<mailto:[email protected]>> wrote:

Dear IPsecme Chairs,

I’m writing to follow up on my note after IETF 122 regarding adoption of 
https://datatracker.ietf.org/doc/draft-dunbar-ipsecme-lightweight-authenticate/ 
. This draft was also presented at the AllDispatch session at IETF 121, where 
it was recommended that IPsecme is the proper venue for the discussion.

The mechanism described in the draft is a key dependency of 
https://datatracker.ietf.org/doc/draft-ietf-rtgwg-multisegment-sdwan/ , which 
is progressing in RTGWG but lacks the IPsec/crypto expertise needed for proper 
review. Without IPsecme involvement, it is difficult to demonstrate that the 
proposed approach has undergone the appropriate expert scrutiny, even though it 
is being referenced and deployed in a routing context.

I read the draft again, and I still don't really understand it. It seems to be 
CPE to Cloud GW but the connection between these is "the internet".

[Linda]  Correct. For example,  an  enterprise has multiple locations, say 
Seattle, Boston, or Singapore.
 Seattle-CPE establishes an IPsec tunnel with Boston-CPE. Instead of best 
effort forwarding of the IPsec encrypted packets from Seattle-CPE to Boston-CPE 
via internet, the IPsec encrypted packets from Seattle-CPE is encapsulated with 
GENEVE header (with Destination address set to the Seattle Cloud GW) to force 
the packet to the Seattle Cloud GW, which in turn can transport the IPsec 
encrypted packets to the Boston Cloud GW via Cloud Backbone. The Boston Cloud 
GW then strip the outer GENEVE header  and forward the original IPsec encrypted 
packets to the Boston-CPE.

This document is about authentication methods to protect the GENEVE header 
(which is not encrypted) between the CPE and its closest Cloud GW.


I don't see how a routing path can (or should) be enforced in such scenarios . 
If it is not, and the path is within the Administrative Domain,

[Linda] It is not about routing path.

then I don't really understand the attack model. This seems to be very similar 
to NASR and other path validation proposals that have not seen
consensus on being a solvable problem. I don't see any relationship to IPsec 
other than that the inner content of the GENEVE packet happens
to be an ESP packet? But that seems orthogonal to the problem and solution 
space. It could also have been a TLS packet for example.
[Linda] the Attack model is about tempering the GENEVE header with the metadata 
for desired forwarding behavior, for example, Excluding specific states, or 
regions.

Additionally, the reasons for not using AH/ESP on the outer layer seems weak. 
We have IPTFS RFC 9347 that should resolve any MTU issues
encountered.
[Linda] since the payload is already encrypted, the document suggests 4 bytes 
HMAC value to avoid MTU issue.

Last, the authentication key is all defined to be "provisioned", which makes 
this not a protocol one simply turns on. And another protocol is
needed to distribute and rotate these keys, and now to deal with streams using 
the just rotated auth key, that is currently unspecified.

[Linda] It is for deployment where there is shared key between the CPE and 
Cloud GW for other traffic (i.e. the services that needs to be terminated 
within the cloud)

Could you advise on the most appropriate path forward to ensure expert review 
by the IPsec community? For example:

  *   Proceed with IPsecme WG adoption (preferred), or

I don't see this being a great for IPsec. It is not using AH, but its own 
(GENEVE?) encapsulation and tweaking some HMAC bits into it)

  *   Pursue AD-sponsored publication in the Security Area with reviewers from 
IPsecme, or

Speaking only for myself, I will not AD sponsor this document.

Paul


Thank you very much.
Linda Dunbar


From: Linda Dunbar
Sent: Tuesday, July 8, 2025 11:28 AM
To: ipsecme-chairs <[email protected]<mailto:[email protected]>>; 
IPSEC <[email protected]<mailto:[email protected]>>
Subject: Request for IPsecme WG Adoption of 
draft-dunbar-ipsecme-lightweight-authenticate

Dear IPsecme WG Chairs,
Thank you again for the opportunity to present 
draft-dunbar-ipsecme-lightweight-authenticate<https://datatracker.ietf.org/doc/draft-dunbar-ipsecme-lightweight-authenticate/>
 at IETF 122. I understand that the current scope of the IPsecme WG may not 
formally cover the specific optimization proposed in this draft.
However, I would like to respectfully request that the IPsecme WG consider 
adopting it as a WG document. The motivation behind this request is that the 
mechanism described in the draft is essential to 
draft-ietf-rtgwg-multisegment-sdwan<https://datatracker.ietf.org/doc/draft-ietf-rtgwg-multisegment-sdwan/>,
 which is actively progressing within the RTGWG. That said, the RTGWG lacks the 
cryptographic and IPsec expertise required to evaluate the proposed lightweight 
authentication solution.
IPsecme, on the other hand, is the natural home for the review and validation 
of IPsec header extensions and authentication mechanisms. Without adoption by 
IPsecme WG, it is difficult to demonstrate that the proposed approach has 
undergone appropriate expert scrutiny, even if it is being referenced and 
deployed in a routing context.
Could you please advise on how best to proceed? I would appreciate your 
guidance on whether this can be adopted or if there is another path forward to 
ensure appropriate review by the IPsecme community.
Warm regards,
Linda Dunbar

_______________________________________________
IPsec mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to