Re: [Idr] Shepherd's report for draft-ietf-idr-bgpls-srv6-ext-05.txt

["Ketan Talaulikar (ketant)" <[e-mail]>]

Hi Sue and Haibo,

Thanks for your review. We've uploaded a new version of the draft to address most of your comments below:

https://www.ietf.org/archive/id/draft-ietf-idr-bgpls-srv6-ext-06.txt

Please check inline for detail responses.

From: Susan Hares <[e-mail]>
Sent: 20 February 2021 01:37
To: [e-mail]
Cc: [e-mail]; Ketan Talaulikar (ketant) <[e-mail]>
Subject: Shepherd's report for draft-ietf-idr-bgpls-srv6-ext-05.txt

Guarav, Ketan, Fils, Mach, Daniel, Bruno:

First of all thank you for your hard work on refining draft-ietf-idr-bgpls-srv6-ext-05.txt.
I'm glad to have this draft in the SRv6 family.

This shepherd report is a combination of my review and a review done by Haibo Wang.   As you recall, I asked for SR-Reviews to review SR drafts in addition to the document shepherds.   I've indicated Haibo's comments in my review.  I appreciate his help in reviewing this IDR draft.

I'm working this weekend in case you want me to review any changes before the draft deadline on Monday.

Cheers, Sue

============
Issues are marked with levels: major or minor
Resolution is marked as: Mandatory or recommended

Issue #1:
========
Location: Section 1:  paragraph 3, sentence
Issue:  Major
Change: Mandatory

Original Text:
/BGP (I-D.ietf.bess-srv6-services)) has been extended to
advertise some of these SRV6 SIDs and SRv6-related information./

Problem description:
The reference to this bess draft means that you must support all the
features within the draft. Specifically, if you are indicating support
for the EVPN sections of this draft you are more likely to hit the
issues described in section 4.9 of draft-ietf-idr-rfc7752bis-05.txt.
[KT] The reference to the BGP SRv6 draft in BESS is informational only. It is not related to section 4.9 of RFC7752-bis since that is applicable only for link-state IGP protocols.

If you wish to keep this text, you have two options:

Option 1) Replace RFC7752 references with RFC7752bis and make appropriate changes
Or
Option 2) Keep dependency on RFC7752 and change the last paragraph of section 10.

For option 2, I provide you with sample text change below.
This text is just a example.

Old Text/The extensions, specified in this document, do not introduce any new
configuration or monitoring aspects in BGP or BGP-LS other than discussed
in [RFC7752]./

New text:/The extensions, specified in this document, do not introduce any new
configuration, monitoring  or error handle aspects of BGP or BGP-LS other
than those discussed in [RFC7752] and [RFC7752bis].  In some use cases
for topologies that might be used for draft-ietf-bess-srv6-services EVPN,
the error handling in [RFC7752bis] section 4.9 may be necessary to handle
the combination of unreachable IGP nodes, BGP-LS with certain BGP Peer topologies,
and service requirements of VPN applications./
[KT] The BESS BGP SRv6 draft is about extensions to BGP address families that are unrelated to this BGP-LS specification.

===============================================================
Issue 2:
Location: Section 3.1, page 6, figure 2
What changes:  Addition to field to clearly show Reserved fields
Issue: minor
Change: Recommended
Problem: Text unclear without reserve word

Old text/
     0                   1
     0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    | |O|                           |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
/
New text/
     0                   1
     0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    | |O|       Reserved            |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+


[KT] Ack

=========================================

Issue 3:
Location: Section 4.1, page 8, bullet beginning "Endpoint Behavior"
Issue: Major
Change: Mandatory

Problem: section 9 is security considerations in
[draft-ietf-spring-srv6-network-programming]
Perhaps you mean section 4 (SR EndPoint descriptions) or
section 10.2 or 10.2.1 with the Initial registrations for
SRv6 Endpoint Behaviors.

Old Text:
/End point Behavior: 2 octet field.  The Endpoint Behavior  code
Point for this SRV6 SID is defined in section 9.2 of
[I-D.ietf-spring-srv6-network-programming]
/
New Text /End point Behavior: 2 octet field.  The Endpoint Behavior  code
Point for this SRV6 SID is defined in section 10.2 of
[I-D.ietf-spring-srv6-network-programming]
/

[KT] Ack

=======================
Issue 4:
Location: Section 4.2, figure 4, page 8
Type: minor
Status: Recommended

Old-Text/
     0 1 2 3 4 5 6 7
    +-+-+-+-+-+-+-+-+
    |B|S|P|         |
    +-+-+-+-+-+-+-+-+
/

New-Text/
     0 1 2 3 4 5 6 7
    +-+-+-+-+-+-+-+-+
    |B|S|P| Reserved|
    +-+-+-+-+-+-+-+-+
/

[KT] Ack

=============
Issue 5:
Location: section 4.2, page 9, paragraph 1
Status: Major
Change: Mandatory to improve text

Reviewers: [Haibo Wang and Susan Hares]
 Old/Text chapter 4.2 SRv6 LAN End.X SID TLV.

   For a LAN interface, normally an IGP node only announces its
   adjacency to the IS-IS pseudo-node (or the equivalent OSPF DR).  The
   SRv6 LAN End.X SID TLV allows a node to announce SRv6 SID
   corresponding to behaviors like END.X
   [I-D.ietf-spring-srv6-network-programming] for its adjacencies to all
   other (i.e. non-DIS or non-DR) nodes attached to the LAN in a single
   instance of the BGP-LS Link NLRI.  Without this TLV, multiple BGP-LS
   Link NLRI would need to be originated for each additional adjacency
   in order to advertise the SRv6 End.X SID TLVs for these neighbor
   adjacencies.
/
[KT] For the record, the text is identical for what we have out for publication for BGP-LS for SR-MPLS : https://datatracker.ietf.org/doc/html/draft-ietf-idr-bgp-ls-segment-routing-ext-16#section-2.2.2

Hares Comments: This paragraph is unclear.  It needs to be rewritten.
[WangHaibo] suggests a way to approach this text:
1.  I suggest to first describe the behavior of the LAN End.X SID explicitly,
such as "each LAN End.X SID is used to specify the
cross-connect to one adjacency node attached to the LAN".
Then describe how to encode the LAN End.X SID TLVs with
the Link NLRI corresponding to the adjacency to DIS or DR.
[KT] It is not the purpose of the BGP-LS specification to describe this and hence we have reference to the individual IGP specifications in the very next paragraph.

==========
Issue 6:
Location: section 4.2., p. 10, bullet beginning EndPoint Behavior
Issue: Major
Change: Mandatory
Problem: Section 9 is in the security section.
If you mean section 10.2, then the text change would be
Old Text/section 9.2 of [I-D.ietf-spring-srv6-network-programming]
New text/section 10.2 of [I-D.ietf-spring-srv6-network-programming]/
[KT] Ack
========
Issue 7:
Location: section 4.2, figure 6
Issue: Minor
Problem: figure unclear without reserved words
Change: Recommended
Old text: /



     0 1 2 3 4 5 6 7

    +-+-+-+-+-+-+-+-+

    |B|S|P|         |

    +-+-+-+-+-+-+-+-+
/
New text:/



     0 1 2 3 4 5 6 7

    +-+-+-+-+-+-+-+-+

    |B|S|P| Reserved|

    +-+-+-+-+-+-+-+-+

[KT] Ack

========
Issue 8:
Location: section 5.1, figure 8
Issue: Minor
Problem: figure unclear without reserved words
Change: Recommended
Old text: /



     0 1 2 3 4 5 6 7

    +-+-+-+-+-+-+-+-+

    |D|             |

    +-+-+-+-+-+-+-+-+
New text:/

     0 1 2 3 4 5 6 7

    +-+-+-+-+-+-+-+-+

    |D| Reserved    |

    +-+-+-+-+-+-+-+-+



[KT] Ack


======
Issue 9:
Location: 2. SRv6 SID NLRI
Status: Major
Changes: Some textual change is mandatory
Problem:
    [Wang Habio comments]
   Now the BGP EPE Peer Node info is advertised with SRv6 SID NLRI, it cause[s] some disadvantages compared to SR-MPLE EPE.
   First, the number of NLRIs needed for SRv6 EPE may be more than MPLS EPE. This is because the NLRI's key is SRv6 SID, but for one EPE Peer node, there may be multiple SIDs, such as End.x with PSP, End.x with USD etc.
   Second, with MPLS EPE, for a direct EBGP Peer, only one NLRI is needed to advertise the link and its Peer node SID, link attributes.  But with the current method for SRv6 EPE, at least two NLRIs are needed, one is the SRv6 SID NLRI for the Peer Node SID, the other is a Link NLRI with the End.X SID (the SID value may be the same while need to be advertised in different NLRIs) and link attributes..
  At current stage maybe it is not suitable to change the encoding, but I suggest to give more detail description about the behavior of advertising the SRv6 Peer node SID and the Peer adjacency SID with corresponding NLRIs for a direct peer and for a peer established on loopback.
[KT] I remember all of this has been discussed on the mailer a few times already (and even offline) during the life of the document in the WG.

[Hares] Solution possibilities:
a) provide text in section 6 prior to NLRI format
b) create section in manageability section providing more details

If have questions on this request for clarifying information,
send email to me, Haibo, or the list.
[KT] I am not able to fully understand what is being asked here and I guess we need further discussion. Is there some text that you have in mind that you can propose?

=========
Issue 10:
Status: Major
Change: Mandatory
Location: 7.1, bullet point that starts "Endpoint Behavior"

Old text/section 9.2 of [I-D.ietf-spring-srv6-network-programming]/
New text / section 10.2 of [I-D.ietf-spring-srv6-network-programming]/
[KT] Ack

========
Issue 11: page 17, Figure 13
Problem: reserved fields
Type: Minor
Status: Recommended
Old Text/
Location:
     0 1 2 3 4 5 6 7
    +-+-+-+-+-+-+-+-+
    |B|S|P|         |
    +-+-+-+-+-+-+-+-+

            Figure 13: SRv6 BGP Peer End.X SID TLV Flags Format

/
NewText/
Location:
     0 1 2 3 4 5 6 7
    +-+-+-+-+-+-+-+-+
    |B|S|P| Reserved|
    +-+-+-+-+-+-+-+-+

            Figure 13: SRv6 BGP Peer Node SID TLV Flags Format
/
[KT] Ack

[Hares comment: Clearly specifying Reserved Fields in figure helps the reader.
[KT] Ack

[Wang Haibo comment:]
   [comment]As Figure 13 is about the Flags of SRv6 BGP Peer Node SID TLV, its name may be changed to SRv6 BGP Peer Node SID TLV Flags Format
[KT] We have flags in most TLVs and I don't see the point of prefixing the TLV name with each of them.

==========
Location: Section 12:  Manageability section
Praise:
Thank you for mentioning the sematic or content checking is left to the consumer of the BGP-LS information.
============

Thanks,
Ketan