Re: [secdir] Secdir last call review of draft-ietf-bess-evpn-oam-req-frmwk-04

[Donald Eastlake <[e-mail]>]

Hi Melinda

On Tue, Feb 16, 2021 at 12:19 AM Melinda Shore via Datatracker
<[e-mail]> wrote:
> Reviewer: Melinda Shore
> Review result: Has Nits
>
> This is a very nicely-structured, efficient, well-written document - among the
> most clearly-written that I've read in a few years.

Thanks.

> Nits:  As a minor point, I am really not a fan of using RFC 2119 language for
> informational documents, and in this case it's being used somewhat
> inconsistently (for example, the lowercase "must" in section 4).

I agree that usage should be consistent. I'll capitalize that "must".

As to the general question of using RFC 2119 language in this
document, I guess we will see what the IESG says.

> I'm also a
> bit unclear on what's intended by "must optionally authenticate" and suggest
> that that should be clarified as to whether what's meant is "mandatory to
> implement but optional to use," or "optional to implement" and should probably
> be a "SHOULD" (or a "should").

Of the three points listed in the Security Considerations, I would
agree that the first two should be SHOULD. But the third, preventing
OAM packet leaking, is, I believe, correctly a MUST.

"Optionally authenticate communicating endpoints" refers, in opinion,
to an implementation requirement, not a use requirement.

> Additionally, it may be helpful to provide an
> example or two of how the EVPN OAM channel could be exploited as a DOS vector,
> and to explain what problem is solved by authenticating EVPN endpoints.

One example would be forged communications that overflow the
capability of an OAM endpoint to maintain state.

Thanks,
Donald
===============================
 Donald E. Eastlake 3rd   [phone] (cell)
 2386 Panoramic Circle, Apopka, FL 32703 USA
 [e-mail]