Re: [Lsr] Benjamin Kaduk's Discuss on draft-ietf-isis-mpls-elc-12: (with DISCUSS and COMMENT)

[Peter Psenak <[e-mail]>]

Hi Ben,

please see inline (##PP2)

On 21/05/2020 21:38, Benjamin Kaduk wrote:
> Hi Peter,
> 
> On Thu, May 21, 2020 at 12:05:39PM +0200, Peter Psenak wrote:
>> Benjamin,
>>
>> thanks for review, please see inline (##PP):
>>
>> On 20/05/2020 00:44, Benjamin Kaduk via Datatracker wrote:
>>> Benjamin Kaduk has entered the following ballot position for
>>> draft-ietf-isis-mpls-elc-12: Discuss
>>>
>>> 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-isis-mpls-elc/
>>>
>>>
>>>
>>> ----------------------------------------------------------------------
>>> DISCUSS:
>>> ----------------------------------------------------------------------
>>>
>>> As for other reviewers, many of my comments duplicate those for the OSPF
>>> document; I expect that the analogous responses apply and am fine if
>>> they only appear for one document's review.
>>>
>>> Here, the question I have about normative language applies to the text
>>> in Section 3:
>>>
>>>      When a router propagates a prefix between ISIS levels ([RFC5302], it
>>>      MUST preserve the ELC signaling for this prefix.
>>>
>>> The scenario in question is analogous to the OSPF cross-area case: is
>>> the router propagating the prefix between ISIS levels required to
>>> implement this document; is preservation of the flag value a new
>>> requirement from this document vs. a preexisting property; and is this
>>> document trying to make normative requirements of devices that don't
>>> implement this document?
>>
>> ##PP
>> this is a new requirement and only applies to the routers that support
>> this document. We are not making normative requirements of devices that
>> don't implement this document, we cannot.
>>
>> Maybe we can add that it only applies to the routers that supports this
>> extension:
>>
>> "When a router supporting this extension propagates a prefix between
>> ISIS levels ([RFC5302], it MUST preserve the ELC signaling for this prefix."
>>
>> Would it work?
> 
> That would work, yes.

##PP2
I have suggested some text to clarify ABR and ASBR behavior in my 
response to Alvaro. Please have a look if that is fine with you. I'm not 
going to add the "When a router supporting this extension", as this was 
not necessary according to Alvaro.

> I think what bothers me about the current text is that it feels like a
> global statement that holds universally everywhere (and as such would be
> "scope overreach").  Even to say something like "ELC signaling MUST be
> preserved when a router propagates a prefix between ISIS levels" (which is
> superficially similar) does not bother me very much, since it presumes a
> knowledge of what ELC signaling is (and thus, implementation of this
> document) in a way that "unknown prefix attribute flag 3" does not.  That
> is, we clearly (given your above explanation) don't expect the "unknown
> prefix attribute flag 3" to get propagaged, and only when we know that it's
> the ELC signalling does the requirement kick in, if that makes sense.
> 
> With respect to Alvaro's clarification, your answer for (1) makes sense;
> thanks!  I think Alvaro has offered to help work out what (if any)
> additional text we might want to be sure that the answer to (2) is clear in
> the document.
> 
>>
>>>
>>> Likewise, the ASBR case for cross-protocol redistribution seems to
>>> rather inherently require understanding the semantics of the flags being
>>> translated.
>>
>> similarly to the above I can chnage to:
>>
>> "When a router supporting this extension redistribute a prefix ..."
>>
>>
>>>
>>>
>>> ----------------------------------------------------------------------
>>> COMMENT:
>>> ----------------------------------------------------------------------
>>>
>>> Section 1
>>>
>>> Should we add a sentence at the end of the last paragraph about how
>>> "this document defines a mechanism to signal the ERLD using IS-IS"?
>>
>> not sure I understand, how is described in the body of the document.
> 
> I think I maybe put a little more detail on the motivation for this
> question in the OSPF document's comments.  Basically, it's about parity
> between Abstract and Introduction -- the Introduction says clearly "this
> draft defines a mechanism to signal the ELC using IS-IS", but there is not
> an analogous statement about the ERLD signaling mechanism.  As such, it's
> an editorial matter of style and you should feel free to ignore the comment
> or do what you see fit.

##PP2
I added the below sentence:

"This document defines a mechanism to signal the ERLD using IS-IS"

Similar has been done to OSPF draft.

> 
>>>
>>>      In cases where LSPs are used (e.g., SR-MPLS [RFC8660], it would be
>>>
>>> side note(?): I don't know that SR-MPLS is so popular so as to be
>>> privileged as the only example given for LSP usage.  If we instead
>>> talked about using IGPs to signal labels, this selection would seem less
>>> surprising to me.
>>
>> this document describes the ELC/ERLD capability signaling for SR MPLS.
>> For non SR MPLS cases, thee are existing mechanisms to learn ELC/ERLD.
>>
>> I can replace the text with:
>>
>> "In cases where SR is used with the MPLS Data Plane"
>>
>> Would it work?
> 
> Sure, thanks.

##PP2
  done.
> 
>>
>>>
>>> Section 3
>>>
>>>      unless all of its interfaces are capable of processing ELs.  If a
>>>      router supports ELs on all of its interfaces, it SHOULD set the ELC
>>>      for every local host prefix it advertises in IS-IS.
>>>
>>> Do we want to say anything about (not) advertising the ELC for other
>>> prefixes?
>>
>> Do we have to? I can add "MUST NOT set ELC with for any other prefix",
>> but I find it unneeded.
> 
> This is the "comment" section of the document, so no, you don't have to :)
> I mostly ask because we give guidance on what to do for local host prefixes
> but say nothing about the other case.  That could be taken to imply that we
> don't know what guidance to give, even though someone skilled in the area
> who thinks through what the mechanism is doing will come to the conclusion
> that it makes no sense/is actively harmful to advertise the ELC for other
> prefixes.

##PP2
I would tend to leave it as it is.

> 
>>>
>>> Section 4
>>>
>>> I agree with Roman's comment about code 2 vs TBD2.
>>
>> that has been fixed already.
>>
>>>
>>>      ERLD in the range between 0 to 255.  The scope of the advertisement
>>>      depends on the application.  If a router has multiple interfaces with
>>>
>>> Just to check: w.r.t. "scope", both this document and the OSPF one only
>>> define usage of this MSD type at the scope of a single node, right?  (I
>>> don't see a particular reason to preclude using it at a different
>>> scope.)
>>
>> the scope here means where the information will be flooded - area only
>> or network wide. No such thing as a node scope.
> 
> Oops, sorry for misremembering that part, and thanks for the clarification.
> 
>>
>>>
>>> Section 6
>>>
>>>         - Bit 3 in the Bit Values for Prefix Attribute Flags Sub-TLV
>>>         registry has been assigned to the ELC Flag.  IANA is asked to
>>>
>>> Is there an "IS-IS" in the name of this registry?
>>
>> no the registry name is "Bit Values for Prefix Attribute Flags Sub-TLV".
> 
> Okay, thanks for confirming.
> 
>>
>>>
>>> Section 7
>>>
>>> Should we say anything about considerations for redistributing ELC/ERLD
>>> information at the ASBR with respect to exposing "internal information"
>>> to external parties?
>>
>> why would this be "internal information" and why redistribution would be
>> "external party"? Redistribution between IGPs is predominantly done
>> between IGPs under same administrative domain.
> 
> I don't have any scenario in mind; I just wanted to check if there were any
> significant considerations to mention.  It sounds like there aren't such
> considerations, and thus nothing to mention.
>>
>>>
>>>      This document specifies the ability to advertise additional node
>>>      capabilities using IS-IS and BGP-LS.  As such, the security
>>>      considerations as described in [RFC7981], [RFC7752], [RFC7794],
>>>      [RFC8491], [RFC8662], [I-D.ietf-idr-bgp-ls-segment-routing-ext] and
>>>
>>> RFC 8662's security considerations have a pretty hard dependency on RFC
>>> 6790's security considerations; it might be worth mentioning 6790
>>> directly in this list as well.
>>
>> would not that be implicit when mentioning RFC 8662?
> 
> Yes; this is mostly a question of style.
> 
>>
>>>
>>>      [I-D.ietf-idr-bgp-ls-segment-routing-msd] are applicable to this
>>>      document.
>>>
>>> Could we also have a brief note that the normal IS-IS authentication
>>> mechanisms serve to protect the ELC/ERLD information?
>>
>> do we need to repeat this every time we add a bit in the TLV?
> 
> It's my preference, though not a strict need.
> 
>>>
>>>      Incorrectly setting the E flag during origination, propagation or
>>>      redistribution may lead to black-holing of the traffic on the egress
>>>      node.
>>>
>>> This is what happens when the E flag should not be set but is
>>> erroneously set.  Should we also say what happens if we should set the E
>>> flag but erroneously clear it (e.g., that poor or no load-balancing may
>>> occur)?
>>
>> yes, there is a text there already:
>>
>> "Incorrectly setting of the ERLD value may lead to poor or no
>> load-balancing of the traffic."
> 
> I thought that was about the ERLD value, not the ELC indicator, but trust
> you to know best.


##PP2
sorry, you are right.

Regarding the "E flag being erroneously clear" - I would think that 
"Incorrectly setting the E flag " covers both cases of E bit being set 
when it should not be set and E bit not being set when it should be set.

What about this:

"Incorrectly setting the E flag during origination, propagation or 
redistribution may lead to poor or no load-balancing of the traffic or 
black-holing of the traffic on the egress node."

thanks,
Peter


> 
>>>
>>> Section 8
>>>
>>> I do see the note in the shepherd writeup about the sixth author (thank
>>> you!); if we're already breaking through the 5-author limit, did we
>>> consider making those who "should be considered as co-authors" listed as
>>> co-authors?
>>
>> I'm not fun of this 5 authors rule to be honest.
>>
>>>
>>> Section 10.1
>>>
>>> Should we reference RFC 7981 from Section 4 as well?  Right now we seem
>>> to only use it for the security considerations, which is not necessarily
>>> enough to qualify it as a normative reference.
>>
>> we reference RFC8491, which references RFC 7981. I don't see a need to
>> reference RFC 7981 directly.
> 
> Okay.
> 
> Thanks for the clarifications and updates!
> 
> -Ben
> 
>