Re: [Lsr] Spencer Dawkins' No Objection on draft-ietf-ospf-ospfv3-segment-routing-extensions-20: (with COMMENT)

[Spencer Dawkins at IETF <[e-mail]>]

Hi, Acee,
On Tue, Dec 4, 2018 at 6:37 PM Acee Lindem (acee) <[e-mail]> wrote:

> Hi Spencer,
>
> I'm replying as document shepherd.
>

It's a pleasure to be talking when we're not both sleepwalking on a 777 :-)

Please note that all of these are comments, so covered under "do the right
thing".


> On 12/4/18, 1:40 PM, "Spencer Dawkins" <[e-mail]>
> wrote:
>
>     Spencer Dawkins has entered the following ballot position for
>     draft-ietf-ospf-ospfv3-segment-routing-extensions-20: No Objection
>
>     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/iesg/statement/discuss-criteria.html
>     for more information about IESG DISCUSS and COMMENT positions.
>
>
>     The document, along with other ballot positions, can be found here:
>
> https://datatracker.ietf.org/doc/draft-ietf-ospf-ospfv3-segment-routing-extensions/
>
>
>
>     ----------------------------------------------------------------------
>     COMMENT:
>     ----------------------------------------------------------------------
>
>     The Introduction would have been much clearer for me if these
> paragraphs were
>     much closer to the top of the section - they're at the bottom of the
> section
>     now.
>
>       This draft describes the OSPFv3 extensions required for Segment
>        Routing with MPLS data plane.
>
>        Segment Routing architecture is described in [RFC8402].
>
>        Segment Routing use cases are described in [RFC7855].
>
>     With that change, I'm not sure how much of the discussion in the
> Introduction
>     would still be required, but do the right thing, of course.
>
>     I'd make the same suggestion for the Abstract,
>
>       Segment Routing (SR) allows a flexible definition of end-to-end paths
>        within IGP topologies by encoding paths as sequences of topological
>        sub-paths, called "segments".  These segments are advertised by the
>        link-state routing protocols (IS-IS and OSPF).
>
>        This draft describes the OSPFv3 extensions required for Segment
>        Routing with MPLS data plane.
>
>     if it was more than two paragraphs long ...
>
> You mean "were" since this is subjective. I'm not sure what you're asking
> for since your comment has something to do with ordering and, as you note,
> the abstract is two paragraphs long.
>

Sorry this wasn't clear.

What I meant was, the Introduction is long enough that moving the
high-order bits to the top is helpful; the Abstract also has the high-order
bits at the bottom, but it's a short distance to the bottom. If you flipped
the Abstract, that might be helpful, and would match the Introduction, but
if you don't, I think making the change in the Introduction would be
sufficient.

>
>     I am thinking that the reference
>
>       There are additional segment types, e.g., Binding SID defined in
>        [RFC8402].
>
>     would be more useful at the beginning of Section 3, because that's
> where you
>     list the additional segment types, but the reference is back in the
>     Introduction (with only one example of the segment types).
>
> Actually, the Binding SID is no longer in the encodings so this could be
> removed.
>

An even better reason to remove this sentence :D ...

That would put the reference to RFC 8402 in Section 3, I assume.


>     I'm thinking the SHOULD in this text
>
>       Existing security extensions as described in [RFC5340] and [RFC8362]
>        apply to these segment routing extensions.  While OSPFv3 is under a
>        single administrative domain, there can be deployments where
>        potential attackers have access to one or more networks in the
> OSPFv3
>        routing domain.  In these deployments, stronger authentication
>        mechanisms such as those specified in [RFC4552] or [RFC7166] SHOULD
>        be used.
>
>     is not an RFC 2119 SHOULD that describes interworking, so something
> like
>
>        In these deployments, stronger authentication
>        mechanisms such as those specified in [RFC4552] or [RFC7166] are
>        needed.
>
> I'll defer to our AD, Alvaro. We have normative text in other "Security
> Considerations" sections.
>

Oh, sure. That wasn't my heartburn at all. My point was

>
>     would be better, but if this IS a SHOULD, I'm curious why you wouldn't
> use
>     stronger authentication mechanisms if they're needed. You might want
> to add
>     guidance about that, so it's not confused with MUST (BUT WE KNOW YOU
> WON'T) as
>     defined in https://tools.ietf.org/html/rfc6919#section-1.
>

that I'm reading the text as saying "you're more vulnerable to attackers,
so you SHOULD use stronger authentication mechanisms, but you might not,
for reasons left to the implementer". Is there a reason that you might
decide not to use stronger authentication mechanisms when you're more
vulnerable to attackers? If so, you might provide it as an example, so the
implementers can do the right thing.

(I spent enough time in the SIP community talking to product managers who
wanted to pay for MUSTs, but didn't think they needed to pay for SHOULDs,
that I'm perhaps overreacting to a problem you folks in RTG don't have. Do
the right thing, of course!)


>     Is there another document that says things like
>
>       Implementations MUST assure that malformed TLV and Sub-TLV defined in
>        this document are detected and do not provide a vulnerability for
>        attackers to crash the OSPFv3 router or routing process.  Reception
>        of a malformed TLV or Sub-TLV SHOULD be counted and/or logged for
>        further analysis.  Logging of malformed TLVs and Sub-TLVs SHOULD be
>        rate-limited to prevent a Denial of Service (DoS) attack
> (distributed
>        or otherwise) from overloading the OSPFv3 control plane.
>
>     ? This doesn't seem very SR-specific, although I'm guessing. If
> there's a
>     broader document, I don't object to including this guidance here, but
> adding a
>     reference to a broader document might be useful.
>
> We do have similar text in section 5 of RFC8362. However, it is not in the
> "Security Considerations" and the statement about rate-limiting is not
> there. It doesn’t hurt to repeat it and it provides confidence that
> "security" has been appropriately "considered".
>

Agree, and thanks for considering all my comments.

Spencer


>
> Thanks,
> Acee
>
>
>
>