Re: [secdir] SECDIR review of draft-ietf-pals-seamless-vccv-02

[Stewart Bryant <[e-mail]>]

On 26/04/2016 15:57, Phillip Hallam-Baker wrote:
> On Tue, Apr 26, 2016 at 10:26 AM, Carlos Pignataro (cpignata)
> <[e-mail]> wrote:
>> Phillip,
>>
>> Many thanks for your review.
>>
>> As you rightly call out, this is indeed an incremental addition — I might add for emphasis a very incremental change.
>>
>> One point of clarification, however, is that this solution as defined does _not_ use BGP. The relevant control protocols’ security considerations are addressed in RFC 5085. This is not 'IPsec pixy-dust' — if you follow the pointers, you will get to the control connection (endpoint and message) security as well as protection for data plane spoofing.
>
> With respect, I disagree.
>
> A collection of pointers to a dozen other documents is not a security
> architecture.
>
> I am aware that this is not BGP which is a layer 3 switching protocol.
I don't think anyone would describe BGP as a layer 3 switching protocol!

It exchanges reachability information, it does not switch packets.

> This is layer 2 but the same security concerns apply. The fact that we
> have seen nation state actors use BGP injection attacks as tools of
> war demonstrate that this is a real concern.

I think you need to explain the attack scenario you have in mind.
>> In re-reading the Security Considerations section (thanks again for the review), I do believe there is an area of improvement: from RFC 5885, since these PWs specify single-hop adjacencies, the document ought to specify the use of GTSM for the IP/UDP encapsulations.
>>
>> I’ll be happy to add that in. Please let me know if you have any concerns with it.
> For an infrastructure of this scale, the security architecture should
> really be described in a separate document and at length.

Let's start with the MPLS variant. What do you think needs to be added 
to RFC5920, and what of that is explicit to PALS?

- Stewart