Re: [pim] [Bier] AD Review of draft-ietf-bier-pim-signaling-11

["Jeffrey (Zhaohui) Zhang" <[e-mail]>]

Hi Alvaro,

About this particular point:

-------------
What is the advantage/difference between using the specification in
this document and unicast-tunneling the PIM messages to the
corresponding node (EBBR)?
-------------

Indeed any tunnel could be used. But since the BIER domain can already do BIER forwarding from IBBR to EBBR (essentially a single p2p tunnel instantiated by BIER), it's natural to just use BIER tunneling.

Jeffrey

-----Original Message-----
From: BIER <[e-mail]> On Behalf Of Alvaro Retana
Sent: Monday, June 21, 2021 3:27 PM
To: [e-mail]
Cc: Nabeel Cocker <[e-mail]>; BIER WG <[e-mail]>; [e-mail]; BIER WG Chairs <[e-mail]>
Subject: [Bier] AD Review of draft-ietf-bier-pim-signaling-11

[External Email. Be cautious of content]


Dear authors:

Thank you for your work and the patience through the publication process.


I just finished reading this document -- it still needs work!   In
general, tightening up the specification is needed -- this is a
Standards Track document!   Also, as written, this document is not
compliant with rfc7761 -- I included most of my concerns about this
point in §3.1.3.1 (search for line 338).


There are some process issues (listed in increasing severity) that I
need the Chairs'/Shepherd's input on:

(1) 6 authors are listed in the front page, but rfc7322 recommends a limit of
    5.  Chairs: Can you please provide justification for going over the
    limit?  [In general, I think that 6 is an ok number -- we just
need to cross the T's...]


(2) I didn't see a specific request from the Chairs asking the pim WG to
    review, or them being cc'd in any of the WGLCs.  Did I miss it?  Given
    the amount of PIM content in this document, that interaction should have
    already happened.

    I realize that there's a significant overlap in participation between the
    two WGs, so I'm cc'ing pim on this message and will forward the IETF LC
    when we get to that stage.


(3) This document replaces draft-hfa-bier-pim-signaling, but that linkage had
    not been indicated in the datatracker when Publication was requested.
    The main issue with this oversight is that draft-hfa-bier-pim-signaling
    has two IPR declarations [1].  The Shepherd's writeup says that no IPR had
    been filed against this document.  Also, I didn't find an IPR poll in the
    archives.

    I have added the link between this document and
    draft-hfa-bier-pim-signaling.

    To address this issue, I need the Chairs to please poll the authors about
    any non-declared IPR.  Please cc the WG (in case anyone else needs to
    declare something) and include a link to the existing disclosures.  Once
    each author has individually replied, I need the Shepherd to update the
    writeup.

    [1]
    https://urldefense.com/v3/__https://datatracker.ietf.org/ipr/search/?submit=draft&id=draft-ietf-bier-pim-signaling__;!!NEt6yMaO-gk!QPgYqq_FW8LdwOUZ7Qy_KSeewvhvngOmUl0Bu8bqj24_0Sj85U11_KrfsA7aYjue$



Finally, I have a general question about the need/applicability for
this work.  This is just for my own education:

What is the advantage/difference between using the specification in
this document and unicast-tunneling the PIM messages to the
corresponding node (EBBR)?  I see the discovery of the EBBR as one of
the differences, but that discovery is outside the scope of this
document -- IOW, the same procedures could be used to discover the
target node without the additional BIER machinery.

Again, I'm just curious.  I suspect other reviewers may have similar questions.


Thanks!

Alvaro.



[Line numbers from idnits.]


...
18      Abstract

[nit] The First two paragraphs could be merged to provide continuity.


20         Consider large networks deploying traditional PIM multicast service.
21         Typically, each portion of these large networks have their own
22         mandates and requirements.

24         It might be desirable to deploy BIER technology in some part of these
25         networks to replace traditional PIM services.  In such cases
26         downstream PIM states need to be signaled over BIER Domain toward the
27         source.

[nit] s/over BIER Domain/over the BIER Domain


29         This draft explains the procedure to signal PIM joins and prunes
30         through a BIER Domain, as such enable provisioning of traditional PIM
31         services through a BIER Domain.

[minor] s/explains/specifies


[minor] s/ joins and prunes / Join/Prune messages /g

Please make sure, throughout the document, that the rfc7761
terminology is followed.


[nit] s/as such enable/enabling the


...
98      1.  Introduction

100        Consider large networks deploying traditional PIM multicast service.
101        Typically, each portion of these large networks have their own
102        mandates and requirements.

[minor] The Introduction is an exact copy of the Abstract.  That in
itself is not necessarily a bad thing.  However, you might want to
take the time here to maybe explain about what the different "mandates
and requirements" are that would result in needing to signal PIM
through the BIER domain.  [This is related to my question at the top
about the need/applicability.]


...
113     2.  Conventions used in this document

115        The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
116        "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
117        document are to be interpreted as describe in [RFC2119].

[major] Please use the template from rfc8174.


119     2.1.  Definitions

121        Some of the terminology specified in [RFC8279] is replicated here and
122        extended by necessary definitions:

[major] Please don't replicate the terminology -- this way we can
avoid the terms falling out of sync or being defined/understood
differently.  Instead, make it clear what the expectations are:

Suggestion>
   An understanding of the BIER architecture [RFC8279] and the related
   terminology is expected.

New terms are ok in this section.  Extensions not so much because they
may change the definition of a term in rfc8279, but a clarification is
ok.


...
153        BBR:

155        BIER Boundary router.  A router between the PIM domain and BIER
156        domain.  Maintains PIM adjacency for all routers attached to it on
157        the PIM domain and terminates the PIM adjacency toward the BIER
158        domain.

160        IBBR:

162        Ingress BIER Boundary Router.  An ingress router from signaling point
163        of view.  It maintains PIM adjacency toward the PIM domain and
164        determines if PIM joins and prunes arriving from PIM domain need to
165        be signaled across the BIER domain.  If so it terminates the PIM
166        adjacency toward the BIER domain and signals the PIM joins/prunes
167        through the BIER core.

169        EBBR:

171        Egress BIER Boundary Router.  An egress router in BIER domain from
172        signaling point of view.  It terminates the BIER packet and forwards
173        the signaled joins and prunes into PIM Domain.

[major] I would have expected for the IBBR/EBBR to be defined as
specific types of a BBR, but the definition doesn't do so.  For
example, the BBR "Maintains PIM adjacency for all routers", as does
the IBBR, but the EBBR doesn't.  IOW, it looks like the EBBR is not a
BBR.   Please review and update the definitions.


...
193     3.  PIM Signaling Through BIER domain
...
208        As per figure 1, the procedures of PIM signaling is done at the BIER
209        boundary router.  The BIER boundary routers (BBR) are connected to
210        PIM capable routers toward the PIM domain and BIER routers toward the
211        BIER domain.  PIM routers in PIM domain continue to send PIM state
212        messages to the BBR.  The BBR will create PIM adjacency between all
213        the PIM routers attached to it on the PIM domain.  That said the BBR
214        does not propagate all PIM packets natively into the BIER domain.
215        Instead when it determines that the PIM join or prune messages needs
216        to be signaled through the BIER domain it will tunnel the PIM packet
217        through the BIER network.  This tunneling is only done for signaling
218        purposes and not for creating a PIM adjacency between the two
219        disjoint PIM domains through the BIER domain.

[] Suggestion>
   Figure 1 illustrates the operation of the BIER Boundary router (BBR).
   BBRs are connected to PIM routers in the PIM domain and BIER routers in
   the BIER domain.  Each BBR determines if a PIM Join or Prune message
   needs to be signaled through the BIER domain and tunnels it through the
   BIER network if so.  This tunneling is only done for signaling purposes
   and not for creating a PIM adjacency between the two disjoint PIM domains
   through the BIER domain.


221        The terminology ingress BBR (IBBR) and egress BBR (EBBR) are relative
222        from signaling point of view.

[minor] s/are relative from signaling/is relative only from a signaling


224        The ingress BBR will determine if an arriving PIM join or prune needs
225        to be signaled across the BIER domain.  While the egress BBR will
226        determine if the arriving BIER packet is a signaling packet and if so
227        it will generate a PIM join/prune packet toward its attached PIM
228        domain.

[minor] The first sentence is redundant with the text above.


230        The BFER and BFIR are BBR from datapath point of view.  It should be
231        noted the new procedures in this draft are only applicable to
232        signaling and there are no changes from datapath point of view.

[minor] I think that mixing the term BBR (defined in this document for
signaling purposes) with BFIR/BFER (at this point) may cause
unnecessary confusion.

Suggestion>
   The new procedures in this document are only applicable to PIM signaling.
   There are no changes to the datapath.


234     3.1.  Ingress BBR procedure

236        IBBR will create PIM adjacency to all PIM routers attached to it
237        toward the PIM domain.

[] Suggestion>
   The IBBR maintains a PIM adjacency [RFC7761] with any PIM router attached
   to it on the PIM domain.


239        When a PIM join or prune for certain (S,G) arrives, the IBBR first
240        determines whether the join or prune is meant for a source that is
241        reachable through the BIER domain.  As an example, this source is
242        located in a disjoint PIM domain that is reachable through the BIER
243        domain.  If so the IBBR will try to resolve the source via an EBBR
244        closest to the source.

246        The procedure to find the EBBR (BFIR from datapath point of view) can
247        be via many mechanisms explained in more detail in upcoming section.

[major] "certain (S,G)"  It is not until §5 that you mention the use
of the mechanisms with ASM.  The text throughout the document feels as
if the operation applies *only* to SSM becasue of the specific used of
"source".  Please mention the applicability earlier so the reader
doesn't have to make assumptions.  Including something in the
Introduction would be ideal -- maybe move a part of §5 there.

Also, some of the text will have to be updated to reflect the use of
the source or RP.


[major] Related.  The mechanisms in this document work to join/leave
the RPT, but not for a source to send Register messages to an RP
across the BIER domain.   Given that the Register messages are
unicast, they should make it to the RP without additional changes,
right?  It would be good if this was explained somewhere.


[major] "mechanisms explained in more detail in upcoming section"
There are examples later on, but not a more detailed explanation.  In
fact, §3.1.1/Appendix A/§3.1.2 declare the EBBR discovery to be out of
scope.  I think that declaring this part of the specification out of
scope is ok, mostly because finding the path towards the source/RP is
a well-known process in PIM.  The difference is that the RPF neighbor
(EBBR in this case) will usually not be directly connected.

I wonder if the process can be described using "RPF", with the caveats
that the RPF neighbor is the EBBR and RPF interface is the BIER tunnel
towards it.

[] Suggestion -- to replace the last two paragraphs (and eliminate
§3.1.1/3,1,2)>

   When a PIM Join or Prune message is received, the IBBR determines whether
   the Source or RP is reachable through the BIER domain.  The EBBR is the
   BFR through which the Source or RP is reachable.  In PIM terms [rfc7761],
   the EBBR is the RPF Neighbor, and the RPF Interface is the BIER "tunnel"
   used to reach it.  The mechanisms used to find the EBBR are outside the
   scope of this document -- some examples are provided in Appendix A.

   If the lookup results in multiple EBBRs, then the EBBR closest to the
   source should be preferred.  The selection algorithm should ensure that
   all signaling for a particular (S,G) or RP is forwarded to a single EBBR.
   For example, the selection can be round robin or use the lowest EBBR IP
   address.


249        After discovering the EBBR and its BFR-ID, the IBBR will include a
250        new PIM Join Attribute in the join/prune message as per [RFC5384].
251        Two new "BIER IBBR" attributes are defined and explained in upcoming
252        section.  The PIM Join Attribute is used on EBBR to obtain necessary
253        BIER information to build its multicast states.  In addition the IBBR
254        will change the PIM signaling packet source IP address to its BIER
255        prefix address (standard PIM procedure).  It will also keep the
256        destination address as the well known multicast IP address.  It then
257        will construct the BIER header.  The signaling packet, in this case
258        the PIM join/prune packet, is encapsulated in the BIER header and
259        transported through BIER domain to EBBR.

[] It took me a while to correlate the "a new PIM Join Attribute in
the join/prune message as per [RFC5384]" with the fact that the
attribute is defined in this document (not rfc5384).

Also, I am not able to find the "two new "BIER IBBR" attributes...in
upcoming section".  I see that the Join Attribute can be either IPv4
or IPv6 -- is that what you meant by 2 attributes?

Suggestion (to replace the first 3 sentences)>
   After discovering the EBBR and its BFR-id, the IBBR MUST use the BIER
   Info Vector (Section 3.1.3) PIM Join Attribute [rfc5384].  The EBBR uses
   this attribute to obtain the necessary BIER information to build its
   multicast state.

   The signaling packet, in this case a PIM Join/Prune message, is
   encapsulated in the BIER Header and forwarded through the BIER domain
   to the EBBR.  The source address of the PIM packets MUST be set to
   belong to the local BFR-prefix.  The destination address MUST be set to
   ALL-PIM-ROUTERS [rfc7761].


There was an interesting suggestion on the list [1] -- I didn't see a reply.

[1] https://urldefense.com/v3/__https://mailarchive.ietf.org/arch/msg/bier/58OIQ_9JbIiUWPrR_Yo7J47GMc8__;!!NEt6yMaO-gk!QPgYqq_FW8LdwOUZ7Qy_KSeewvhvngOmUl0Bu8bqj24_0Sj85U11_KrfsMaUFr9l$


261        The IBBR will track all the PIM interfaces on the attached PIM domain
262        which are interested in a certain (S,G).  It creates multicast states
263        for arriving (S,G)s from PIM domain, with incoming interface as BIER
264        "tunnel" interface and outgoing interface as the PIM domain
265        interface(s) on which PIM Join(s) were received on.

[minor] "arriving (S,G)s"  I guess you mean "arriving Join messages"...


...
288     3.1.3.  PIM Signaling packet construction at IBBR

[nit] s/at IBBR/at the IBBR


[] Let's separate the definition of the new attribute from its use.
IOW. in §3.1.3 specify the attribute (in general -- independent of the
application), and use §3.1.3.1 to indicate the use and packet
construction.

The title in §3.1.3 should then be: The BIER Info Vector PIM Join Attribute


290        To ensure all necessary BIER information needed by EBBR is present in
291        the BIER signaling message, a new PIM Join Attribute [RFC5384] is
292        used.  EBBR can use this attribute to build its multicast states, as
293        described in EBBR procedure section.  This new PIM join Attribute is
294        added to PIM signaling message on the IBBR.  Its format is as follow:

[minor] Note that the name of this attribute is not mentioned until
§7. :-(   BTW, do you want to keep the name as "BIER Info Vector" or
expand it to "BIER Information Vector"?

Suggestion>
   The BIER Info Vector PIM Join Attribute [rfc5384] is defined as follows:


296           0                   1                   2                   3
297           0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
298          +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
299          |F|E|    Type=tbd |     Length  |  Addr Family    | BIER info
300          +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-...

[major] s/Type=tbd/Attr_Type
That is the name in rfc5384.


302                            Figure 2: PIM Join Attribute

304        F bit: The Transitive bit.  Specifies whether this attribute is
305        transitive or non-transitive.  MUST be set to zero.  This attribute
306        is ALWAYS non-transitive.

[] NEW>
   F-bit: Transitive Attribute, as specified in [rfc5384].  MUST be set to
   zero as this attribute is always non-transitive.


[major] What should the receiver do if the F-bit is set?


308        E bit: End-of-Attributes bit.  Specifies whether this attribute is
309        the last.  Set to zero if there are more attributes.  Set to 1 if
310        this is the last attribute.

[] NEW>
   E-bit: End of Attributes, as specified in [rfc5384].


312        Type: TBD assign by IANA.

[] NEW>
   Attr_Type: TBD


314        Length: The length in octets of the attribute value.  MUST be set to
315        the length in octets of the BIER info +1 octet to account for the
316        Address Family field.  For IPv4 AF Length = 7+1 For IPv6 AF Length =
317        19+1.

[] NEW>
   Length: length of the value field, as specified in [rfc5384].  MUST be
   set to the length of the BIER Info field + 1.  For IPv4 the length is 8,
   and 20 for IPv6.


[major] What should a receiver do if the length is not set correctly
with respect to the AF?


319        Addr Family: Signaled PIM Join/Prune address family as defined in
320        [RFC7761].

[] NEW>
   Addr Family: PIM address family, as specified in [rfc7761].


[major] What should a receiver do if the received AF is not IPv4 or IPv6?


322        BIER Info: IBBR Prefix (IPv4 or IPv6), SD, bfr-id as per below figure

[minor] Move the definition of the fields to be after the figure.

NEW>
   BIER Info:  This field is described in Figure 3.


324           0                   1                   2                   3
325           0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
326          +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
327          ~                 IBBR Prefix IPv4 or IPv6                      ~
328          +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
329          | subdomain-id  |        BFR-ID                 |
330          +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

332                         Figure 3: PIM Join Attribute detail

[major] What is the IBBR prefix?  I guess it is the IBBR's BFR-prefix.


[major] What should the receiver do if the BFR-prefix is not routable?


[major] s/subdomain-id/sub-domain-id
Use SD in the figure to make sure it fits.


[major] s/BFR-ID/BFR-id/g


[major] What should a receiver do if the BFR-id doesn't match the
BFIR-id in the BIER Header?


[major] What should the receiver do if multiple instances of this
attribute are received?


[major] Are there any considerations related to the interaction with
other PIM Join attributes?  For example the Vector attribute.  I am
asking both from the point of view of the IBBR receiving other
attributes from the PIM domain, *and* the IBBR including/forwarding
them through the BIER domain.  If any of the received attributes are
transitive, should they be forwarded through the BIER domain?


334     3.1.3.1.  BIER packet construction at IBBR

[nit] s/at IBBR/at the IBBR


336        The BIER header will be encoded with the BFR-id of the IBBR(with
337        appropriate bit set in the BitString) and the PIM signaling packet is
338        then encapsulated in the packet.

[minor] I think the whole process can be summarized in a single
paragraph -- no need for the figure or the definitions below.

Suggestion>
   The BIER Header is used according to the specification in [rfc8296] for
   a BIER packet to be sent from the IBBR to the EBBR.  The PIM Join/Prune
   message is encapsulated in it including the BIER Info Vector PIM Join
   Attribute (Section 3.1.3).



[major] Compliance with rfc7761.

Even though it is not explicitly mentioned until §5, I want to talk
about not sending Hellos and the overall compliance with rfc7761 here.

TL;DR: as written, not sending Hello messages is not compliant with rfc7761.


rfc7761/§4.3.1 says this:

   Note that neighbors will not accept Join/Prune or Assert messages
   from a router unless they have first heard a Hello message from that
   router.  Thus, if a router needs to send a Join/Prune or Assert
   message on an interface on which it has not yet sent a Hello message
   with the currently configured IP address, then it MUST immediately
   send the relevant Hello message without waiting for the Hello Timer
   to expire, followed by the Join/Prune or Assert message.

This text requires (*MUST*) a Hello *before* sending a Join/Prune.

Later, in rfc7761/§4.5 the text is a little more lenient (even contradictory):

   In general, a PIM Join/Prune message should only be accepted for
   processing if it comes from a known PIM neighbor.  A PIM router hears
   about PIM neighbors through PIM Hello messages.  If a router receives
   a Join/Prune message from a particular IP source address and it has
   not seen a PIM Hello message from that source address, then the
   Join/Prune message SHOULD be discarded without further processing.

This text recommends (*SHOULD*) discarding a Join/Prune if no Hello
has been received.  This opens a very small door to process the
Join/Prune...

As we all know, a requirement is stronger than a recommendation.  It
is very hard for me to consider how this document can work around not
sending the required Hello *and* still claim to be compliant with
rfc7761.


One possible way forward is to also send PIM Hellos through the BIER
domain.  This would immediately bring this document in compliance with
rfc7761 and remove other concerns (below).


The other option is to not send PIM Hellos through the BIER domain.
There are several points that then have to be addressed:

- What about rfc7761?  Focus on a strong justification to ignore the
recommendation in §4.5.

- What capabilities are expected the BBRs to have?  What are the
effects of those expectations not being met?

- How does the EBBR determine the PIM availability of the IBBR?  IOW,
not using Hellos means that the EBBR cannot determine whether the IBBR
(more explicitly, its PIM functionality) is still available.

- There are other potential security issues.  For example, any node
can send Joins/Prunes with irrelevant information causing traffic to
unnecessarily traverse the network.  This is not a new threat for PIM
(even a known neighbor can do that), but in this case the attack
surface is expanded.



[major] The issue of the MTU of the BIER tunnel was brought up during
WGLC [1], but I didn't see a resolution.  FWIW, I agree with Stig's
observation that if a significant amount of state is to be carried in
the Join/Prune, then it is important to be aware of the effective MTU
through the BIER domain.

There's some text in §4.4.1/rfc7761 that could be repurposed here.

[1] https://urldefense.com/v3/__https://mailarchive.ietf.org/arch/msg/bier/-YBAPXTXOKXfkOqcD36_54F1_Bo__;!!NEt6yMaO-gk!QPgYqq_FW8LdwOUZ7Qy_KSeewvhvngOmUl0Bu8bqj24_0Sj85U11_KrfsOKduR5t$


...
368     3.2.  Signaling PIM through the BIER domain procedure

[minor] This section seems unnecessary as it states the obvious.  It
is ok to leave it in for completeness.  I do think however that the
second paragraph is not needed at all as it just repeats (without
using the exact same language) what is already specified in rfc8279.


370        Throughout the BIER domain the BIER forwarding procedure is on par
371        with [RFC8279].  No BIER router will examine the BIER packet
372        encapsulating the PIM signaling packet.  As such there is no
373        multicast state built in the BIER domain.

[minor] s/on par with /according to


375        The packet will be forwarded through the BIER domain until it reaches
376        the BER with matching BFR-ID as in the BIERHeader.Bitstring.  EBBR
377        will remove the BIER header and examine the PIM IPv4 or IPv6
378        signaling packet further as per EBBR Procedure section.


380     3.3.  EBBR procedure

382        EBBR will remove the BIER header and determine this is a signaling
383        packet.  The Received PIM join/prune Signaling packet is processed as
384        if it were received from neighbors on a virtual interface, (i.e. as
385        if the pim adjacency was present, regardless of the fact that there
386        is no adjacency).

[nit] s/EBBR will remove/The EBBR removes


[minor] s/BIER header/BIER Header/g

388        The EBBR will build a forwarding table for the arriving (S,G) using
389        the obtained BFIR-id and the Sub-Domain information from BIER Header
390        and/or the PIM join Attributes added to the PIM Signaling packet.  In
391        short it tracks all IBBRs interested in this (S,G).  This is
392        explained in section 4.1.

[nit] s/The EBBR will build/The EBBR builds


[major] "forwarding table"  rfc7761 refers to the state process using
a TIB (indirectly referring to a multicast forwarding table); please
use consistent explanations with the existing RFCs.


[minor] "arriving (S,G)"  Maybe something like "membership
information" from the PIM Join/Prune messages.


[minor] s/using the obtained BFIR-id and the Sub-Domain information
from BIER Header and/or the PIM join Attributes added to the PIM
Signaling packet./using the information obtained from the IBBR (in the
BIER Header and the PIM Join/Prune message).


[minor] "explained in section 4.1"   §4.1 is just paragraph.  Instead
of pointing the user there, please move that text here.  However, it
seems to me that the next paragraph is redundant with anything that
§4.1 says, so it may not be needed at all.


394        The multicast state on EBBR will contain PIM domain incoming
395        interfaces, according to PIM specification and outgoing interfaces
396        based on the above procedure to build the forwarding table.

[minor] NEW>
   The multicast state on the EBBR contains interfaces on the local PIM
   domain as incoming interfaces.  The outgoing interface is set to the BIER
   "tunnel" used to reach the IBBR.


398        It should be noted EBBR will maintain PIM adjacency toward the PIM
399        domain and all PIM routers which are connected to it.  At this point
400        the end-to-end multicast traffic flow setup is complete.

[] Suggestion>
   The EBBR maintains a PIM adjacency [RFC7761] with any PIM router attached
   to it on the PIM domain.  At this point the end-to-end multicast traffic
   flow setup is complete.


...
413     4.2.  Datapath traffic flow

415        When the multicast data traffic arrives on the BFIR (EBBR) the router
416        will find all the interested BFERs for that specific (S,G).  The
417        router then constructs the BIERHeader.BitString with all the BFER
418        interested in the group and will forward the packet to the BIER
419        domain.  The BFER(s) will accept the packets and remove the BIER
420        header and forward the multicast packet as per pre-built multicast
421        state for (S,G) and its outgoing interfaces.

[] NEW>
   When multicast data traffic arrives on the BFIR (EBBR) it forwards,
   through the BIER domain, the traffic to all interested IBBRs following
   the procedures specified in [RFC8279].   The BFER(s) (IBBR(s)) also
   follow the procedures in [rfc8279] and forward the multicast packet
   through its outgoing interface(s).


423     5.  PIM-SM behavior

425        The procedures described in this document can work with Any-Source
426        Mutlicast (ASM) as long as static Rendevous Point (RP) or embedded RP
427        for IPv6 is used.  Future drafts would cover Bootstrap Router (BSR)
428        and more complicated SM discovery mechanisms.

[minor] s/can work with/can be used with


[nit] s/as static/as a static


[major] Please add a Normative reference to rfc3956.


[minor] The last sentence is not needed, please remove it.


430        It should be noted that this draft only signals PIM Joins and Prunes
431        through the BIER domain and not any other PIM message types including
432        PIM Hellos or Asserts.  As such functionality related to these other
433        type of massages will not be possible through a BIER domain with this
434        draft and future drafts might cover these scenarios.  As an example
435        DR selection should be done in the PIM domain or if the PIM routers
436        attached to IBBRs are performing DR selection there needs to be a
437        dedicated PIM interface between these routers.

[major] As mentioned before, not including Hellos may cause some
issues (including not conforming to rfc7761).  It is not clear to me
that the justification ("such functionality related to these other
type of massages will not be possible") applies to Hellos.  What about
the capabilities, what is expected?  How does the IBBR know if the
EBBR even supports Join attributes??


[minor] The example doesn't apply to sending messages over the BIER domain.


439        In case of PIM ASM Static RP or embedded RP for IPv6 the procedure
440        for leaves joining RP is the same as above.  It should be noted that
441        for ASM, the EBBRs are determined with respect to the RP instead of
442        the source.

[major] Instead of including this explanation, update the
specification part of this document to take into account the ASM case.


444     6.  Applicability to MVPN

[] This section starts by saying that "With just minor changes, the
above procedures apply to MVPN", but the description goes into more
that "minor changes" (including a potential "more complicated label
allocation scheme") and points at other details not properly
specified.  If you want to include the MVPN case, then this section
will need more details, references, etc.

Personally I would suggest removing it.


...
478     7.  IANA Considerations

480        In the "PIM Join Attribute Types" registry, IANA to assign a new
481        value [TBD] to the BIER Info Vector.

[major] NEW>
   IANA is requested to assign a value (TBD) to the BIER Info Vector PIM
   Join Attribute from the PIM Join Attribute Types registry.


483     8.  Security Considerations

485        The procedures of this document do not, in themselves, provide
486        privacy, integrity, or authentication for the control plane or the
487        data plane.  For a discussion of the security considerations
488        regarding the use of BIER, please see [RFC8279] and [RFC8296].
489        Security considerations regarding PIM protocol is based on [RFC7761].

[nit] s/[last sentence]/The security considerations for PIM [rfc7761]
also apply.


[major] Some of the risks introduced in this document are not new
(forging a Join/Prune, for example), but can now be exploited in a
different way (across a BIER domain).  The effect may be more
significant as it may result in traffic forwarded through the network
without need; think congestion, use of resources, etc.  The Prune may
have the exact opposite effect.

Note that you don't need to provide a solution, just mention that as
in rfc7761 forging is an issue (but with the BIER domain increased
scope).


[major] I mentioned receiving a Join/Prune without a neighbor before.
Note that rfc7761 recomments this in §6.2:

   A PIM router SHOULD provide an option to limit the set of neighbors
   from which it will accept Join/Prune, Assert, and Hello messages.
   Either static configuration of IP addresses or an IPsec security
   association MAY be used.  Furthermore, a PIM router SHOULD NOT accept
   protocol messages from a router from which it has not yet received a
   valid Hello message.

The recommendation to limit the set of neighbors should be highlighted
in this document.  In fact, I think it should be required.


...
496     10.  References

498     10.1.  Normative References

500        [RFC4607]  "H. Holbrook, B. Cain, "Source-Specific multicast for
501                   IP"", October 2016.

[major] Unused reference.


...
507     10.2.  Informative References
...
513        [RFC2119]  "S. Brandner, "Key words for use in RFCs to Indicate
514                   Requirement Levels"", March 1997.

[major] This reference should be Normative.


516        [RFC5384]  "A. Boers, I. Wijnands, E. Rosen, "PIM Join Attribute
517                   Format"", November 2008.

[major] This reference should be Normative.


...
526        [RFC7761]  "B.Fenner, M.Handley, H. Holbrook, I. Kouvelas, R. Parekh,
527                   Z.Zhang "PIM Sparse Mode"", March 2016.

[major] This reference should be Normative.


529        [RFC8296]  "IJ. Wijnands, E. Rosen, A. Dolganow, J. Yantsura, S.
530                   Aldrin, I. Meilik, "Encapsulation for BIER"", January
531                   2018.

[major] This reference should be Normative.


533        [RFC8401]  "Ginsberg, L., Przygienda, T., Aldrin, S., and Z. Zhang,
534                   "BIER Support via ISIS"", June 2018.

[major] Unused reference.


536        [RFC8444]  "Psenak, P., Kumar, N., Wijnands, IJ., Dolganow, A.,
537                   Przygienda, T., Zhang, Z., and S. Aldrin, "OSPF Extensions
538                   for Bit Index Explicit Replication"", June 2018.

[major] Unused reference.


540        [RFC8556]  "Rosen, E., Ed., Sivakumar, M., Wijnands, IJ., Aldrin,
541                   S.,Dolganow, A., and T. Przygienda, "Multicast VPN Using
542                   BIER"", March 2018.

[major] Unused reference.


544     Appendix A.

546        This appendix provides some examples and routing procedures that can
547        be used to determine the EBBR on IBBR.  It should be noted, the PIM
548        domains can be either part of the same IGP area as BIER domain(single
549        area) or are stitched to the BIER domain via an ABR or ASBR routers.
550        As such on IBBR, there can be many different procedures to determine
551        the EBBR.  Not all procedures are listed below.

[nit] s/examples and routing procedures/examples of routing procedures


[nit] s/EBBR on IBBR/EBBR at the IBBR


[major] I think that talking about overlaps (or lack of) between PIM,
BIER, and general flooding domains may raise significant questions
related to the specification.  Personally, I wouldn't go there.


553     A.1.  SPF

[major] SPF may be an understood reference to a link-state protocol,
but it may not be properly understood by the casual reader.  Please
mention link-state protocols instead.


...
570     A.2.1.  Static Route

572        On IBBR there can be a static route configured for the source, with
573        source next-hop set as EBBR BIER prefix.

[nit] NEW>
   A static route to the source can be configured on the IBBR with the
   next-hop set as the EBBR's BFR-prefix.


575     A.2.2.  Interior Border Gateway Protocol (iBGP)

577        Consider the following topology:

579                               BBR                   BBR
580                               EBBR                  IBBR
581                 |--PIM Domain--|-----BIER domain-----|--PIM domain--|
582            S--( A )----------( B ) ---- ( C ) ---- ( D )----------( E )--h

[nit] Using "BBR" on top of EBBR/IBBR is redundant.


584                               Figure 5: Static Route

586        Suppose BGP is enable between EBBR (B) and IBBR (D) and the PIM
587        Domain routes are redistributed to the BIER domain via BGP.  This
588        would include the Multicast Source IP address (S), which resides in
589        the PIM Domain.  In such case BGP should use the same loopback
590        interface as its next-hop as the BBR prefix.  This will ensure that
591        all PIM domain routes, including the Multicast Source IP address (S)
592        are resolve via BBR's BIER prefix id as their next-hop.  When the
593        host (h) triggers a PIM join message to IBBR (D), IBBR tries to
594        resolve (S).  It resolves (S) via BGP installed route and realizes
595        its next-hop is EBBR (B).  IBBR will use this next-hop (B) to find
596        its corresponding BIER bit index.

[minor] "BGP should use the same loopback interface as its next-hop as
the BBR prefix"   There is no relationship between BGP and the
BFR-prefix.

Suggestions>
   Figure 5 shows BGP enabled between the EBBR (B) and the IBBR (D), and
   it is used to distribute the routes from the the PIM Domain through the
   BIER domain, including the Multicast Source IP address (S).  The peering
   address used for BGP should belong to the same block as the BFR-prefix if
   the EBBR  to ensure that all routes from the PIM domain are resolved via
   the EBBR's BFR-prefix as their next-hop.


598        This procedure is inline with [RFC6826] mLDP in-band signaling
599        section

[] I don't see the relevance of this sentence.  Please take it out.

[End of Review -11]

_______________________________________________
BIER mailing list
[e-mail]
https://urldefense.com/v3/__https://www.ietf.org/mailman/listinfo/bier__;!!NEt6yMaO-gk!QPgYqq_FW8LdwOUZ7Qy_KSeewvhvngOmUl0Bu8bqj24_0Sj85U11_KrfsBfjvnBa$

Juniper Business Use Only