Re: [secdir] Secdir last call review of draft-ietf-detnet-mpls-05

[Stewart Bryant <[e-mail]>]

> On 12 Mar 2020, at 02:29, Watson Ladd <[e-mail]> wrote:
> 
> I don't see any reason why RFC 3552's guidelines shoudn't apply to this draft.

No you mis the point.

The underlay (MPLS) is resilient to the level required by RFC3552 so that really is out of scope because we are not describing MPLS itself but what in MPLS terminology is an application running over MPLS.

As far as I can see the RFC3552 view of security was the necessary security to be applied to a best effort service.

The types of attack that very few people have experience of is dynamic attacks. The only people that have experience of them, and I am sure you know, are people doing time transfer. Now I think that this is largely an unsolved problem, particularly with fine grained timing. I have worked on systems where this has accidentally occurred (two 1588 streams with different clocks beating), but this sets a floor on the lower deterministic delay bound and as I noted earlier DN is not looking for min delay, just a delay floor.

However that is not really a security problem, it is an operational performance problem.

> 
> If there is a MPLS exception I'd like to see the rules that should be applied.
> 
> As is I don't see any reason why the assumptions can't be explicitly
> spelled out, either in the Security Considerations or elsewhere.

We can certainly note that we assume that DN over MPLS will be run over a well managed MPLS network if that helps.

If you have explicit suggestions for text we will look at them.

- Stewart