Paul,
Thank you for your review and feedback.

The situation that I see this experimental method being useful in is now
explained more explicitly in new Section 1.5, and to me draws close
analogies to both IP network access and the email validation of RFC 8823.

For the IP access case, there are interior transit networks and edge access
networks and through delegation of IP address space edge networks can
obtain blocks of addresses for which they have been delegated authority to
control. It is up to each edge network to determine how to authorize use of
any edge node address, which might involve things like IEEE 802.1x
Port-based Network Access Control or 802.11i Wifi Protected Access. But any
interior transit routers don't see those details, they only trust that the
edge network has properly controlled the use of its allocated addresses.

Similarly, the means by which an email edge message transfer agent (MTA)
authenticates a message user agent (MUA) and authorizes access to a
particular address is assumed and invisible to interior MTAs relaying
messages from and to that address. There are plenty of mechanisms to do
that access control via SMTP, POP, IMAP and their derivatives. Any interior
MTAs must trust that the edge MTA has properly controlled the use of its
mail domain. And there are mechanisms, such as DKIM of RFC 6376, for edge
MTAs to vouch for its users to those interior MTAs.

For BPv7 networks, sets of valid Node IDs can be similarly delegated to
edge networks and left to each edge to determine how to authenticate and
authorize edge nodes. This could include authentication and authorization
at any under layer, including the same types of access control discussed
above for IP network access or methods such as use of TLS within the
TCPCLv4 of RFC 9174 which allows IP-based or DNS-name-based authentication
and authorization to be used. Once an edge node is authorized to be able to
access an edge network using a specific identity, the edge routers may add
additional security to satisfy transit or interior network needs (which is
discussed in Section 4 of the draft). But the analogy to IP and email
address access still holds, I believe.

All of that is separate from the on-path impersonation threat described in
Section 6.2 of the draft, and maybe that section needs more explanation
similar to the above. Even in cases where there was no integrity gateway
and an interior network required no in-band, cryptographic integrity
controls I think this experimental method is secure to on-path attack. The
main requirement which mitigates an on-path attack is in Section 3.3 of the
draft, which is that the Challenge Bundle from the ACME server shall have
primary and payload integrity traced to the ACME server's BP node and
trusted by the node-under-validation. This constraint might not be clear in
the current text, and additional discussion in Section 4 or 6.2 could
improve this.

Even if a client's BP Agent ignored the security requirements on the
Challenge Bundle, an on-path attacker would still not have access to the
client's account key thumbprint or the server-supplied `token-chal` both
necessary to generate the corresponding key authorization digest. If the
attacker waits until the response bundle is sent and intercepts it, the
worst the attacker can do is to drop the bundle and prevent the validation
from succeeding. Any alteration there would cause the key auth. digest to
fail to match and cause the validation to fail.

Does any of this help out? Or suggest that more explanation (but not a
change of behavior) is needed?
Brian S.

On Wed, Jun 4, 2025 at 9:01 PM Paul Wouters via Datatracker <
[email protected]> wrote:

> Paul Wouters has entered the following ballot position for
> draft-ietf-acme-dtnnodeid-18: Abstain
>
> When responding, please keep the subject line intact and reply to all
> email addresses included in the To and CC lines. (Feel free to cut this
> introductory paragraph, however.)
>
>
> Please refer to
> https://www.ietf.org/about/groups/iesg/statements/handling-ballot-positions/
> for more information about how to handle DISCUSS and COMMENT positions.
>
>
> The document, along with other ballot positions, can be found here:
> https://datatracker.ietf.org/doc/draft-ietf-acme-dtnnodeid/
>
>
>
> ----------------------------------------------------------------------
> COMMENT:
> ----------------------------------------------------------------------
>
> I am filing a ballot of Abstain, as I can't fully comprehend the security
> model
> of ACME issuing endpoint certificates through an insecure Bundle Protocol
> based
> network which is not end-to-end secure. The ACME protocol security is
> bootstrapped by any ACME client being able to know that its content is only
> readable by the ACME server, and visa versa. But that is not the case here.
>
>         It is possible for intermediate BP nodes to
>         encapsulate-and-encrypt Challenge and/or Response Bundles while
>         they traverse untrusted networks, but that is a DTN configuration
>         matter outside of the scope of this document.
>
> Defining the lack of this core ACME requirement as out-of-scope kind of
> confuses me as to what security guarantees ACME issued certificates offer.
> Any
> Intermediate BP node can get ACME certs issued for Node IDs behind it. So
> what
> does an ACME Node ID certificate really mean from a security point of view?
>
>
>
>
_______________________________________________
Acme mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to