Hi Jon,

On Mon, Feb 11, 2019 at 06:57:55PM +0000, Jonathan Hardwick wrote:
> Hi Benjamin
> 
> As I was preparing the next revision of this document, I realised I'd 
> forgotten to reply to your email.  (I thought I had replied, but I can find 
> no evidence that I did so!)

I have been guilty of the same more times than I can count; what matters is
that we get the work done, really.

> Anyway, I think we have converged.  Please see my proposals below, and 
> apologies for my tardiness.

Indeed we have converged.

> Cheers
> Jon
> 
> ----
> > Isn't the 'F' bit fully redundant with NT?
> > 
> > [Jon] The F bit determines whether NAI is appended, or not.  NT gives the 
> > type of the NAI that is appended.  The NT field has no meaning if F=0.
> 
> My (implicit) proposal was to assign a value for the NT field that means "no 
> NAI is present", and to treat that value of NT as meaning that F=0, with any 
> other value of NT meaning F=1, in which case the F bit does not add any new 
> information.
> 
> [Jon2] Understood. Unfortunately, we have to make a pragmatic decision here 
> as implementations are already fielded.  I have changed the text to say this 
> to clarify.
> 
> NEW
>    NAI Type (NT):  Indicates the type and format of the NAI contained in
>       the object body, if any is present.  If the F bit is set to zero
>       (see below) then the NT field has no meaning and MUST be ignored
>       by the receiver.
> END NEW

Thanks.  (It is wholly unsurprising that making actual wire-level changes
here is to be avoided unless absolutely necessary, and this definitely does
not qualify as absolutely necessary.)

> ----
> > It's interesting that the routing protocols' MSD value supersedes the 
> > PCE-obtained one, given that all three (IS-IS, OSPF, and BGP-LS) documents 
> > have text to the effect of "PCEP provides a way to get this, but if PCEP is 
> > not used you can use our thing", which a naive reader would take to mean 
> > that PCEP is intended to be the primary distribution mechanism.
> > 
> > [Jon] Probably needs to be fixed in the other documents.  PCEP is a fairly 
> > blunt tool for specifying the MSD, which it can only do per node.  The IGPs 
> > can do better by specifying the MSD per network interface.
> 
> Ah.  Unfortunately, at least the IS-IS document is already published as an 
> RFC (8491) and thus immutable.  (I didn't check the others.)  I don't know 
> whether it would be appropriate to add this explanation into this document 
> since the other documents cannot (all) be changed.
> 
> [Jon2] I've reworded the text to make the relationship between PCEP and the 
> IGPs clearer:
> 
> NEW
>    Note that the MSD value exchanged via the SR-PCE-CAPABILITY sub-TLV
>    indicates the SID/label imposition limit for the PCC node.  It is
>    anticipated that, in many deployments, the PCCs will have network
>    interfaces that are homogeneous with respect to MSD (that is, each
>    interface has the same MSD).  In such cases, having a per-node MSD on
>    the PCEP session is sufficient; the PCE SHOULD interpret this to mean
>    that all network interfaces on the PCC have the given MSD.  However,
>    the PCE MAY also learn a per-node MSD and a per-interface MSD from
>    the routing protocols, as specified in: [RFC8491]; [RFC8476];
>    [I-D.ietf-idr-bgp-ls-segment-routing-msd].  If the PCE learns the
>    per-node MSD of a PCC from a routing protocol, then it MUST ignore
>    the per-node MSD value in the SR-PCE-CAPABILITY sub-TLV and use the
>    per-node MSD learned from the routing protocol instead.  If the PCE
>    learns the MSD of a network interface on a PCC from a routing
>    protocol, then it MUST use the per-interface MSD instead of the MSD
>    value in the SR-PCE-CAPABILITY sub-TLV when it computes a path that
>    uses that interface.
> END NEW

That is very helpful, at least to me; thank you.

> 
> ----
> > I don't think it's a good idea to just refer to "the security 
> > mechanisms of [RFC 5440] and [RFC8281]", since there are qualitative 
> > differences between the TCP authentication schemes and the full-on 
> > encryption provided by TLS and IPsec.  (Also, what exactly are the 
> > security mechanisms of RFC8281 supposed to be -- a quick look only 
> > sees the new guidance to apply resource
> > limits?)  The second paragraph only requires integrity protection to avoid 
> > the vulnerability mentioned there, but the third paragraph requires 
> > confidentiality protection to preent the attack.
> > 
> > [Jon] RFC 8281 pulls in RFC 8231 in its security section, as you note 
> > above.  RFC 8231 says
> > 
> >    As a general precaution, it is RECOMMENDED that these PCEP extensions
> >    only be activated on authenticated and encrypted sessions across PCEs
> >    and PCCs belonging to the same administrative authority, using
> >    Transport Layer Security (TLS) [PCEPS], as per the recommendations
> >    and best current practices in [RFC7525].
> 
> That still doesn't seem to give the reader much guidance about which cases 
> require encryption and which cases can safely proceed with just integrity 
> protection (even though I'm happy to see the general recommendation to use 
> TLS!).
> 
> [Jon2] OK, I understand. The strong steer is that TLS should be used all the 
> time, but we haven't mandated it, so you are right, more guidance should be 
> given.  I think your analysis is correct, so I've changed the text in the 
> second and third paragraphs as follows.
> 
> NEW
>    Note that this specification enables a network controller to
>    instantiate a path in the network without the use of a hop-by-hop
>    signaling protocol (such as RSVP-TE).  This creates an additional
>    vulnerability if the security mechanisms of [RFC5440], [RFC8231] and
>    [RFC8281] are not used.  If there is no integrity protection on the
>    session, then an attacker could create a path which is not subjected
>    to the further verification checks that would be performed by the
>    signaling protocol.
> 
>    Note that this specification adds the MSD field to the OPEN message
>    (see Section 4.1.2) which discloses how many MPLS labels the sender
>    can push onto packets that it forwards into the network.  If the
>    security mechanisms of [RFC8231] and [RFC8281] are not used with
>    strong encryption, then an attacker could use this new field to gain
>    intelligence about the capabilities of the edge devices in the
>    network.
> END NEW

Thanks!

-Benjamin

_______________________________________________
Pce mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/pce

Reply via email to