<?xml version='1.0' encoding='utf-8'?>
<!DOCTYPE rfc SYSTEM "rfc2629.dtd" [
<!ENTITY RFC7854 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.7854.xml">
<!ENTITY RFC7432 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.7432.xml">
<!ENTITY RFC9135 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.9135.xml">
<!ENTITY RFC9972 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.9972.xml">
<!ENTITY RFC2119 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.2119.xml">
<!ENTITY RFC8174 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.8174.xml">
<!ENTITY RFC5925 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.5925.xml">
<!ENTITY I-D.ietf-bess-evpn-yang SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml3/reference.I-D.ietf-bess-evpn-yang.xml">
<!ENTITY I-D.ietf-grow-bmp-yang SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml3/reference.I-D.ietf-grow-bmp-yang.xml">
<!ENTITY I-D.ietf-grow-bmp-stats-informational-tlv SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml3/reference.I-D.ietf-grow-bmp-stats-informational-tlv.xml">
<!ENTITY I-D.wang-grow-bmp-bgp-rib-stats-ext SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml3/reference.I-D.wang-grow-bmp-bgp-rib-stats-ext.xml">
<!ENTITY I-D.zhuang-grow-bmp-enhancement-for-vrf-loc-rib SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml3/reference.I-D.zhuang-grow-bmp-enhancement-for-vrf-loc-rib.xml">
<!ENTITY I-D.dikshit-grow-bmp-rd-scoped-rib-stats SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml3/reference.I-D.dikshit-grow-bmp-rd-scoped-rib-stats.xml">
<!ENTITY RFC4684 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.4684.xml">
<!ENTITY RFC9069 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.9069.xml">
]>
<rfc submissionType="IETF" docName="draft-saum-grow-bmp-afi-safi-evpn-09" category="std" ipr="trust200902">
	<!-- Generated by id2xml 1.6.0 on 2026-11-19T10:59:52Z -->
	<?rfc strict="yes"?>
	<?rfc compact="yes"?>
	<?rfc subcompact="no"?>
	<?rfc symrefs="yes"?>
	<?rfc sortrefs="no"?>
	<?rfc text-list-symbols="*-o+"?>
	<?rfc toc="yes"?>
	<front>
	<title abbrev="EVPN-Specific BMP RIB Stats">EVPN-Specific BMP RIB Statistics Extensions</title>
	<author initials="S." surname="Dikshit" fullname="Saumya Dikshit">
	<organization>Hewlett Packard Enterprise</organization>
	<address><postal>
	<street>Mahadevpura</street>
	<city>Bangalore</city>
	<region>Karnataka</region>
	<code>560 048</code>
	<country>India</country>
	</postal>
	<email>saumya.dikshit@hpe.com</email>
	</address>
	</author>

	<author initials="M." surname="Srivastava" fullname="Mukul Srivastava">
	<organization>Hewlett Packard Enterprise</organization>
	<address><postal>
	<street>10 Technology Park Dr</street>
	<city>Westford</city>
	<region>MA</region>
	<code>01886</code>
	<country>United States of America</country>
	</postal>
	<email>mukul.srivastava@hpe.com</email>
	</address>
	</author>

	<author initials="C." surname="Lin" fullname="Changwang Lin">
	<organization>New H3C Technologies</organization>
	<address><postal>
	<street>8 Yongjia North Road</street>
	<city>Beijing</city>
	<region>Haidian District</region>
	<code>100094</code>
	<country>China</country>
	</postal>
	<email>linchangwang.04414@h3c.com</email>
	</address>
	</author>

	<date year="2026" month="November" day="20"/>
	<abstract><t>
   This document defines new EVPN-specific BGP Monitoring Protocol
   (BMP) statistics types.  These extensions include scalar counters
   for EVPN route types specifically, while also keeping scope for
   defining counters which are EVPN route-type agnostic but related to
   BGP-EVPN RIB like, number of multihoming Ethernet Segments, number
   of multihomed EVIs, number of aliased paths, number of dynamic
   inter-VRF route leaking (IVRL).</t>

	<t>
   This revision (-06) firms up the specification for working group
   consideration: it adds a Relationship to Other Work section that
   explicitly maps this document to concurrently circulating GROW
   contributions on BMP statistics and TLV extensions and invites
   coordination with their authors, corrects the YANG alignment
   reference, strengthens Security Considerations, and adds a worked
   example. No statistics semantics, Stat Type/Subtype allocations, or
   wire formats defined in -05 are changed.</t>

	</abstract>
	</front>

	<middle>
	<section title="Introduction" anchor="sect-1"><t>
   <xref target="RFC7854"/> defines different BMP statistics message types and
   <xref target="RFC9972"/> extends the BMP statistics to provide more insights into
   the BGP RIBs.  This document extends the BMP RIB statistics by
   adding EVPN-specific statistics types to monitor EVPN-related
   events in Adj-RIB-In, Adj-RIB-Out, and Local-RIB for the EVPN
   address family (AFI=25, SAFI=70) as defined in <xref target="RFC7432"/> and
   extended by <xref target="RFC9135"/> for integrated routing and bridging, forms
   the basis for these statistics.</t>

	<t>
   The format of the BMP statistics message remains same as defined in
   <xref target="RFC7854"/>.</t>

	<section title="Requirements Language" anchor="sect-1.1"><t>
   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY",
   and "OPTIONAL" in this document are to be interpreted as described
   in BCP 14 <xref target="RFC2119"/> <xref target="RFC8174"/> when, and only when, they appear in
   all capitals, as shown here.</t>

	</section>

	</section>

	<section title="EVPN-Specific BMP Statistics Extensions" anchor="sect-2"><t>
   This section defines EVPN-specific statistics types for BMP,
   applicable to the EVPN address family (AFI=25, SAFI=70).  These
   statistics type for all 5 RIB views: Adj-RIB-In-Pre,
   Adj-RIB-In-Post, Local-RIB, Adj-RIB-Out-Pre and Adj-RIB-Out-Pre.
   and to define more specific counters other than the route-types in
   future.</t>

	<section title="Global EVPN Route Statistics" anchor="sect-2.1"><t>
   The section defines global EVPN route or other information
   statistics, e.g., the overall Ethernet Auto-discovery Route
   counters.</t>

	<section title="Statistics Definitions" anchor="sect-2.1.1"><t>
   This section defines different statistics type for global EVPN
   monitoring type.</t>

	<t><list style="symbols"><t>Type=TBD_01: Statistics for global EVPN routes received in
      Adj-RIB-In before inbound policy is applied.</t>

	<t>Type=TBD_02: Statistics for global EVPN routes in Adj-RIB-In
      after inbound policy is applied.</t>

	<t>Type=TBD_03: Statistics for global EVPN routes in the local RIB.</t>

	<t>Type=TBD_04: Statistics for global EVPN routes in Adj-RIB-Out
      before outbound policy is applied.</t>

	<t>Type=TBD_05: Statistics for global EVPN routes in Adj-RIB-Out
      after outbound policy is applied.</t>

	<t>Type=TBD_06: Statistics for global EVPN information of
      ESs/EVIs/aliased paths.</t>

	</list>
	</t>

	</section>

	<section title="Statistics Format" anchor="sect-2.1.2"><t>
   The global statistics are whole-router counts and are not scoped to a
   Route Distinguisher.  They are the non-RD companion to the per-EVI
   statistics of <xref target="sect-2.2"/>, in the sense of the companion
   statistics described in Section 4 of
   <xref target="I-D.dikshit-grow-bmp-rd-scoped-rib-stats"/>.  Each counter
   is encoded in Stat Data (Section 4.8 of <xref target="RFC7854"/>) as a
   1-octet Stats Type, an optional 1-octet Stats Subtype, and a 64-bit
   gauge:</t>

	<figure><artwork><![CDATA[
Global EVPN statistics Stat Data (whole router, no RD):

  Field           Size       Value / notes
  -------------   --------   ----------------------------------
  Stats Type      1 octet    EVPN metric (0 to 4; see below)
  Stats Subtype   1 octet    EVPN route type; only for Type 0/1
  Gauge           8 octets   the count
]]></artwork>
	</figure>
	<t><list style="symbols"><t>Stats Type (1 octet): EVPN stats type.  Defined types are:<list style="symbols"><t>Type = 0: EVPN route statistics in the RIB.</t>

	<t>Type = 1: EVPN leaked route statistics in the local RIB.</t>

	<t>Type = 2: EVPN multihoming Ethernet Segments (ESs)
         statistics.</t>

	<t>Type = 3: EVPN multihomed EVPN instances (EVIs) statistics.</t>

	<t>Type = 4: EVPN aliased paths statistics.</t>

	</list>
	</t>

	<t>Stats Subtype (1 octet): Indicates the EVPN route type
      associated with this statistic, as defined in:<list style="symbols"><t>Subtype = 0: Reserved.</t>

	<t>Subtype = 1: EVPN route type 1 (Ethernet Auto-discovery).</t>

	<t>Subtype = 2: EVPN route type 2 (MAC/IP Advertisement).</t>

	<t>Subtype = 3: EVPN route type 3 (Inclusive Multicast Ethernet
         Tag Route).</t>

	<t>Subtype = 4: EVPN route type 4 (Ethernet Segment Route).</t>

	<t>Subtype = 5: EVPN route type 5 (IP Prefix Route).</t>

	<t>Subtype = 6: EVPN route type 6 (Selective Multicast Ethernet
         Tag Route).</t>

	<t>Subtype = 7: EVPN route type 7 (IGMP Join Synchronization
         Route).</t>

	<t>Subtype = 8: EVPN route type 8 (IGMP Leave Synchronization
         Route).</t>

	<t>Subtype = 9: EVPN route type 9 (Per-Region I-PMSI A-D
         route).</t>

	<t>Subtype = 10: EVPN route type 10 (S-PMSI A-D route).</t>

	<t>Subtype = 11: EVPN route type 11 (Leaf A-D route).</t>

	</list>
	</t>

	</list>
	</t>

	<t>
   Stats Subtype is only used when Stats Type is set to 0 and 1;
   otherwise, Stats Subtype is omitted.  A BMP implementation MUST
   ignore unrecognized stat types on receipt, and likewise MUST ignore
   unexpected data in the Stat Data field.</t>

	</section>

	</section>

	<section title="Per-EVI EVPN Route Statistics" anchor="sect-2.2"><t>
   The section defines specific EVI EVPN route statistics, e.g.,
   per-EVI IP Prefix Route counters.</t>

	<section title="Statistics Definitions" anchor="sect-2.2.1"><t>
   This section defines different statistics type for per-EVI EVPN
   monitoring type.</t>

	<t><list style="symbols"><t>Type=TBD_07: Statistics for per-EVI EVPN routes received in
      Adj-RIB-In before inbound policy is applied.</t>

	<t>Type=TBD_08: Statistics for per-EVI EVPN routes in Adj-RIB-In
      after inbound policy is applied.</t>

	<t>Type=TBD_09: Statistics for per-EVI EVPN routes in the local
      RIB.</t>

	<t>Type=TBD_10: Statistics for per-EVI EVPN routes in Adj-RIB-Out
      before outbound policy is applied.</t>

	<t>Type=TBD_11: Statistics for per-EVI EVPN routes in Adj-RIB-Out
      after outbound policy is applied.</t>

	</list>
	</t>

	<t>
   Reporting the leaked-route, multihoming Ethernet Segment, multihomed
   EVI, and aliased-path statistics (the global Stats Types 1 through 4 of
   <xref target="sect-2.1"/>) per RD, rather than only at the global
   level, is an open item.  Doing so would need either additional Stat
   Type code points or a profile extension, because the container in
   <xref target="sect-2.2.2"/> carries only the RD as its instance key; it
   is noted here as a candidate for a future revision and is not specified
   in this document.  Multihoming statistics are in any case naturally
   scoped by Ethernet Segment Identifier rather than by RD, a different
   axis that this document does not model.</t>

	</section>

	<section title="Statistics Format" anchor="sect-2.2.2"><t>
   Per-EVI EVPN route statistics use the RD-Scoped Statistics Container
   defined in Section 3.1 of
   <xref target="I-D.dikshit-grow-bmp-rd-scoped-rib-stats"/>, with Family
   set to 2 (EVPN), AFI set to 25, and SAFI set to 70.  The 21-octet
   container is carried as the Stat Data of a Stat Type TLV (Section 4.8
   of <xref target="RFC7854"/>); the full encapsulation inside the BMP
   Statistics Report, with a Stat Length of 21, is shown below:</t>

	<figure><artwork><![CDATA[
BMP Stat Type TLV header (Section 4.8 of RFC 7854):

  Field                 Size       EVPN value
  -------------------   --------   -----------------------
  Stat Type             2 octets   per RIB view, TBD_07-11
  Stat Length           2 octets   21

RD-Scoped Statistics Container (21 octets), Family = EVPN:

  Field                 Size       EVPN value
  -------------------   --------   -----------------------
  Family                1 octet    2 (EVPN)
  Subtype               1 octet    EVPN route type, 1-11
  AFI                   2 octets   25
  SAFI                  1 octet    70
  Route Distinguisher   8 octets   source EVI's RD
  Gauge                 8 octets   route count

Fields are packed in the order shown, with no padding; Stat Length = 21
counts Family through Gauge.
]]></artwork>
	</figure>
	<t><list style="symbols"><t>Family (1 octet): EVPN, assigned from the "BMP RD-Scoped
      Statistics Family" registry of
      <xref target="I-D.dikshit-grow-bmp-rd-scoped-rib-stats"/>.</t>

	<t>Subtype (1 octet): the EVPN route type (1 to 11), as listed for
      Stats Subtype in <xref target="sect-2.1.2"/>.</t>

	<t>AFI (2 octets), SAFI (1 octet): 25 and 70 for EVPN.  Family and
      AFI/SAFI are two different axes and are both carried; see
      Section 3.1 of
      <xref target="I-D.dikshit-grow-bmp-rd-scoped-rib-stats"/> for the
      rationale.  For EVPN the two coincide.</t>

	<t>Route Distinguisher (8 octets): the source EVI's RD.  The field
      MUST be populated with the corresponding source EVI's RD.  The RD
      identifies the instance as it appears in the monitored RIB.  In the
      Local-RIB case (Stats Type TBD_09) a locally configured EVI
      typically has a single RD, so the gauge is a per-EVI count.  In the
      Adj-RIB-In and Adj-RIB-Out cases (Stats Types TBD_07, TBD_08,
      TBD_10, and TBD_11), and in particular on a Route Reflector or Route
      Server, a single EVI that spans several originating PEs can appear
      under several RDs; the gauge is then a per-RD slice, and recovering a
      per-EVI total requires correlating those RDs by an attribute such as
      a Route Target, which the reporting node does not necessarily
      hold.</t>

	<t>Gauge (64 bits): the EVPN route count.</t>

	</list>
	</t>

	<t>
   In RFC 7432 <xref target="RFC7432"/> VLAN-aware bundle services, several
   broadcast domains share a single MAC-VRF and a single RD and are
   distinguished only by the Ethernet Tag ID carried in the NLRI.
   Because the container keys a counter on the RD and the Subtype (route
   type) alone, a per-EVI gauge for such a service aggregates all
   broadcast domains in the bundle.  An implementation reporting
   VLAN-aware bundles MUST treat the gauge as the whole-bundle value.
   Adding an Ethernet Tag ID selector to report individual broadcast
   domains is an open item for a future revision or profile.</t>

	<t>
   Under EVPN Integrated Routing and Bridging <xref target="RFC9135"/>, a
   single Type 2 (MAC/IP) route can be imported into both a MAC-VRF and an
   IP-VRF.  A counter using this container counts each route once under the
   RD it carries in the monitored RIB; an implementation MUST NOT count
   the same NLRI more than once under one RD.  Where a MAC-VRF and an
   IP-VRF on the same node share one RD, the Subtype (EVPN route type)
   distinguishes their routes, so the Type 2 or Type 3 routes of the
   MAC-VRF and the Type 5 routes of the IP-VRF remain separable in the
   report even when the RD is identical.</t>

	</section>

	</section>

	<section title="Worked Example" anchor="sect-2.3"><t>
   This non-normative example shows a per-EVI IP-Prefix route count in the
   Local-RIB, carried as the RD-Scoped Statistics Container, for EVI Route
   Distinguisher 65001:100 reporting 42 active EVPN route type 5
   prefixes:</t>

	<figure><artwork><![CDATA[
Statistics Report
  Per-Peer Header: Loc-RIB Instance Peer (RFC 9069)
  Stat TLV
    Stat Type = TBD_09 (loc-rib-evpn-route-per-evi-stats)
    Stat Len  = 21
    Stat Data (RD-Scoped container):
      Family  = 2    (EVPN)
      Subtype = 5    (EVPN route type 5, IP-Prefix)
      AFI     = 25   SAFI = 70
      RD      = 65001:100
      Gauge   = 42
  => 42 EVPN type-5 routes in EVI 65001:100 (Local-RIB)
]]></artwork>
	</figure>
	<t>
   A collector receiving this Stat Data, together with the per-EVI Remote
   VRF Information carried by a companion mechanism such as
   <xref target="I-D.zhuang-grow-bmp-enhancement-for-vrf-loc-rib"/> (see <xref target="sect-3"/>),
   can attribute the 42 routes both to a count (this document) and, where
   available, to the specific remote VRF/EVI instance they were learned
   from (that document's TLV).</t>

	</section>

	</section>

	<section title="Collector Deployment Topologies and RD Correlation" anchor="sect-topology"><t>
   Because an EVPN Route Distinguisher ensures route uniqueness rather
   than serving as a global tenant identifier, how a collector aggregates
   per-RD counts depends on where the BMP session terminates.</t>

	<figure><artwork><![CDATA[
Case A: Direct PE / VTEP sessions (Loc-RIB)

   VTEP1 (RD1) --BMP--+
   VTEP2 (RD2) --BMP--+--> Collector
   VTEP3 (RD3) --BMP--+

   One local RD maps to one EVI; the per-RD gauge is the per-EVI count.

Case B: Shared Route Reflector session (Adj-RIB-In)

   VTEP1 (RD1) --+
   VTEP2 (RD2) --+--> RR --BMP--> Collector
   VTEP3 (RD3) --+

   One EVI spans N VTEPs with auto-derived RDs, so the RR exports N
   per-RD slices; the collector sums the EVI by Route Target.
]]></artwork>
	</figure>
	<t><list style="symbols"><t>Direct PE or VTEP sessions (Local-RIB): each PE exports its local
      EVI under a single local RD, and the per-RD gauge maps one-to-one
      to the EVI.</t>

	<t>Shared Route Reflector sessions (Adj-RIB-In): when an EVI spans
      several VTEPs using auto-derived RDs (for example a VTEP loopback
      and a VNI), the Route Reflector exports one per-RD slice per VTEP.
      Reconstructing an EVI-wide total requires correlating those RDs by
      their Route Target, which a collector obtains from route monitoring
      rather than from the counter itself, since the Route Reflector does
      not hold the EVI-to-Route-Target state.</t>

	</list>
	</t>

	</section>

	<section title="Relationship to Other Work and Call for Collaboration" anchor="sect-3"><t>
   Several documents currently active in GROW define statistics or
   TLV extensions to BMP that intersect with the scope of this
   document.  This section maps this document explicitly against that
   work and records specific collaboration proposals; it does not
   change the semantics or wire format of any statistic defined in
   <xref target="sect-2"/>.</t>

	<t><list style="symbols"><t><xref target="I-D.zhuang-grow-bmp-enhancement-for-vrf-loc-rib"/> defines a
      Remote VRF Information TLV that lets a BMP Loc-RIB [RFC9069]
      report identify the remote VRF/EVI a VPN route was learned
      from.  That document's <xref target="sect-1"/> example is structurally
      identical to the dynamic inter-VRF route leaking (IVRL)
      scenario this document's Type=1 (EVPN leaked route statistics)
      counter is intended to quantify: their TLV supplies the "from which remote instance" topology detail, while our Type=1/Type=3
      counters supply the "how many, and how often" volumetric
      detail.  Today, an operator has no single-document way to
      correlate the two.  We propose to the authors of that document
      (S. Zhuang, N. Geng, H. Wang) that a joint follow-up either (a)
      adds an EVPN-scoped counterpart to their Remote VRF Information
      TLV referencing this document's per-EVI statistics types, or
      (b) is captured by generalizing this document's per-EVI Stat
      Data format, as outlined in
      <xref target="I-D.dikshit-grow-bmp-rd-scoped-rib-stats"/>.  We welcome
      co-authorship or contributed text on this point.</t>

	<t><xref target="I-D.wang-grow-bmp-bgp-rib-stats-ext"/> introduces a 4-bit Flags
      field (Triggered/Differential bits) into the Stat Type TLV
      header used for its fine-grained filtering statistics.  The
      per-EVI and global EVPN counters defined in Section 2 of this
      document use the unmodified <xref target="RFC7854"/>/<xref target="RFC9972"/> Stat Data
      layout and do not currently have an equivalent
      triggered/differential reporting capability.  Rather than this
      document defining a second, incompatible flags convention, we
      propose that the Flags field defined in
      <xref target="I-D.wang-grow-bmp-bgp-rib-stats-ext"/> <xref target="sect-2"/> be generalized
      (in that document, or in a shared companion document) so that
      EVPN statistics types defined here can also set the
      Triggered/Differential bits when a future revision of this
      document adopts them.  We invite L. Wang, N. Geng, L. Li, and S.
      Zhuang to discuss whether that flags octet should be lifted out
      into its own short, reusable BMP Stat Type TLV extension
      document that both efforts can reference.</t>

	<t><xref target="I-D.ietf-grow-bmp-stats-informational-tlv"/>, already adopted by
      the GROW working group and co-authored by two of this document's
      authors (M. Srivastava, C. Lin) together with S. Kolenchery,
      defines a Statistics Information TLV carrying minimum, maximum,
      average, and median values alongside a gauge.  The per-EVI and
      global EVPN gauges defined in Section 2 of this document are
      directly applicable inputs to that TLV; implementations
      reporting EVPN statistics SHOULD consider also applying
      <xref target="I-D.ietf-grow-bmp-stats-informational-tlv"/> to smooth
      short-term EVI-level churn (e.g., during mass Ethernet Segment
      failover) for collector consumption.</t>

	<t><xref target="I-D.dikshit-grow-bmp-rd-scoped-rib-stats"/>, authored by the same
      authors as this document, proposes generalizing the Stat Data
      format of <xref target="sect-2.2"/> (Stats Type, Stats Subtype, 8-octet RD,
      Gauge) into a family-agnostic "RD-Scoped Statistics" container
      applicable beyond EVPN to L3VPN and MVPN.  A future revision of
      this document may adopt that container as the wire format for
      <xref target="sect-2.2"/>, registering EVPN as Family=2 in that document's
      registry, without changing any Stat Type allocation requested in
      Section 5 of this document.</t>

	</list>
	</t>

	<t>
   Authors of the documents referenced above are encouraged to bring
   any of these coordination points to the GROW mailing list or to a
   future GROW session so that Stat Type TLV conventions and registry
   entries can be aligned before multiple, mutually incompatible
   variants reach working group last call.</t>

	</section>

	<section title="YANG Model Alignment" anchor="sect-4"><t>
   The counters defined in this document are intended to align with a
   YANG model for EVPN monitoring.  The -05 revision of this document
   referenced <xref target="I-D.ietf-bess-evpn-yang"/> for that alignment; that
   document has since expired (it was last updated in March 2019) and
   SHOULD NOT be relied upon as a current reference.  Route type
   counters are intended to map to YANG leaves for EVPN route
   statistics, multihoming and IVRL counters are intended to align
   with Ethernet Segment and VRF-related leaves, and segment failure
   impact is intended to align with failure event counters, but no
   currently maintained EVPN YANG module implements these leaves.</t>

	<t>
   Separately, <xref target="I-D.ietf-grow-bmp-yang"/>, the active GROW working group
   YANG model for BMP itself, does not yet model <xref target="RFC9972"/>'s advanced
   statistics types or the EVPN-specific statistics types defined in
   this document.  The authors intend to propose a YANG augmentation
   of <xref target="I-D.ietf-grow-bmp-yang"/> covering both, and, if
   <xref target="I-D.dikshit-grow-bmp-rd-scoped-rib-stats"/> proceeds, to model that
   document's generalized container instead of a one-off EVPN-specific
   augmentation.  Implementations SHOULD treat both YANG alignment
   statements in this section as informative and subject to revision
   as that augmentation work proceeds.</t>

	</section>

	<section title="IANA Considerations" anchor="sect-5"><t>
   This document requests that IANA assign new BMP statistics types
   from the "BMP Statistics Type" registry for the following:</t>

	<t><list style="symbols"><t>Type TBD_01: rib-in-pre-evpn-route-stats: Statistics for global
      EVPN routes received before inbound policy is applied.</t>

	<t>Type TBD_02: rib-in-post-evpn-route-stats: Statistics for
      global EVPN routes after inbound policy is applied.</t>

	<t>Type TBD_03: loc-rib-evpn-route-stats: Statistics for global
      EVPN routes in the local RIB.</t>

	<t>Type TBD_04: rib-out-pre-evpn-route-stats: Statistics for
      global EVPN routes before outbound policy is applied.</t>

	<t>Type TBD_05: rib-out-post-evpn-route-stats: Statistics for
      global EVPN routes after outbound policy is applied.</t>

	<t>Type TBD_06: rib-out-post-evpn-info-stats: Statistics for
      global EVPN information of ESs/EVIs/aliased paths.</t>

	<t>Type TBD_07: rib-in-pre-evpn-route-per-evi-stats: Statistics
      for per-EVI EVPN routes received before inbound policy is
      applied.</t>

	<t>Type TBD_08: rib-in-post-evpn-route-per-evi-stats: Statistics
      for per-EVI EVPN routes after inbound policy is applied.</t>

	<t>Type TBD_09: loc-rib-evpn-route-per-evi-stats: Statistics for
      per-EVI EVPN routes in the local RIB.</t>

	<t>Type TBD_10: rib-out-pre-evpn-route-per-evi-stats: Statistics
      for per-EVI EVPN routes before outbound policy is applied.</t>

	<t>Type TBD_11: rib-out-post-evpn-route-per-evi-stats: Statistics
      for per-EVI EVPN routes after outbound policy is applied.</t>

	</list>
	</t>

	<t>
   The 11 types are sufficient to provide generic monitoring of EVPN
   route types across Adj-RIB-In, Adj-RIB-Out, and Local-RIB.</t>

	</section>

	<section title="Operational Considerations" anchor="sect-op"><t>
   The per-EVI statistics in <xref target="sect-2.2"/> are reported per RD,
   so the number of Stat Data instances grows with the number of
   monitored EVIs and RDs.  A producer SHOULD be able to rate-limit the
   reporting interval and SHOULD allow an operator to select a subset of
   EVIs, RDs, route types (the Subtype), or Stat Types, so that a
   high-cardinality fabric does not emit an unbounded volume of
   statistics.  To bound the volume further, a producer MAY report only
   the counters that changed since the previous report, for example by
   layering a triggered or differential reporting mechanism on a Stat
   Type that uses this format.</t>

	<t>
   When these statistics are exported from an Adj-RIB-In or Adj-RIB-Out
   on a Route Reflector or Route Server, the reported counts reflect only
   the routes present in that RIB after inbound or outbound policy, Route
   Target Constrain <xref target="RFC4684"/>, and best-path or Add-Path
   selection have been applied.  Such counts describe the constrained or
   selected subset seen by that node, not the entire fabric, and counts
   for the same RD from two different nodes are not directly comparable
   unless those nodes observe the same set of routes.  Reconstructing
   EVI-wide totals across a fabric is addressed in
   <xref target="sect-topology"/>.</t>

	</section>

	<section title="Security Considerations" anchor="sect-6"><t>
   The security considerations of <xref target="RFC7854"/> apply to this document.</t>

	<figure><artwork><![CDATA[
The per-EVI statistics defined in Section 2.2 reveal, per Route
Distinguisher, the volume of EVPN routes, leaked routes, and
multihoming state associated with a specific EVI.  Where an EVI
corresponds to a customer-facing tenant instance, this is
customer-instance-level operational detail comparable in
sensitivity to the remote-VRF topology information discussed in
the Security Considerations of
[I-D.zhuang-grow-bmp-enhancement-for-vrf-loc-rib].  As with that
document, implementations SHOULD restrict BMP session access to
authorized and trusted monitoring devices and SHOULD use transport
security (e.g., TCP-AO [RFC5925] or a TLS profile for BMP)
consistent with Section 11 of [RFC7854].  No new protocol
interactions beyond the BMP Statistics Report message defined in
[RFC7854] are introduced by this document.
]]></artwork>
	</figure>
	</section>

	<section title="Contributors" anchor="sect-7"><t>
   The following people made significant contributions to this
   document:</t>

	<t>
   To be added.</t>

	</section>

	<section title="Acknowledgements" anchor="sect-8"><t>
   The authors would like to acknowledge the GROW working group and
   the authors of the related documents cited in <xref target="sect-3"/> for their
   ongoing work in this space; specific review credit will be added
   as discussion on the coordination points in <xref target="sect-3"/> progresses.</t>

	</section>

	</middle>

	<back>
	<references title="Normative References">
	&RFC7854;
	&RFC7432;
	&RFC9135;
	&RFC9972;
	&RFC2119;
	&RFC8174;
	&RFC5925;
	&I-D.dikshit-grow-bmp-rd-scoped-rib-stats;
	</references>
	<references title="Informative References">
	&I-D.ietf-bess-evpn-yang;
	&I-D.ietf-grow-bmp-yang;
	&I-D.ietf-grow-bmp-stats-informational-tlv;
	&I-D.wang-grow-bmp-bgp-rib-stats-ext;
	&I-D.zhuang-grow-bmp-enhancement-for-vrf-loc-rib;
	&RFC4684;
	&RFC9069;
	</references>
	<section title="Changes Between Revisions" anchor="sect-changes"><t>
   This section is to be removed before publishing as an RFC.</t>

	<t>
   v08 -&gt; v09</t>

	<t><list style="symbols"><t>Expanded the <xref target="sect-2.2.2"/> field table to show the full
      encapsulation of the container inside the BMP Statistics Report
      Stat Type TLV, with the EVPN binding Family=2, AFI=25, SAFI=70 and
      a Stat Length of 21.</t>

	<t>Presented the global statistics format
      (<xref target="sect-2.1.2"/>) as a field table and framed it as the
      non-RD, whole-router companion to the per-EVI container, aligned
      with Section 4 of
      <xref target="I-D.dikshit-grow-bmp-rd-scoped-rib-stats"/>.</t>

	<t>Added a Collector Deployment Topologies and RD Correlation section
      (<xref target="sect-topology"/>) contrasting direct PE or VTEP
      sessions, where one local RD maps to one EVI, with a shared Route
      Reflector session, where an EVI spans several auto-derived RDs and
      the collector reconstructs the EVI total by Route Target.</t>

	<t>Extended the Operational Considerations (<xref target="sect-op"/>)
      to cover operator subset selection by route type and
      triggered or differential reporting, and to cross-reference the new
      topology section.</t>

	</list>
	</t>

	<t>
   v07 -&gt; v08</t>

	<t><list style="symbols"><t>Per-EVI EVPN route statistics (<xref target="sect-2.2.2"/>) now use
      the RD-Scoped Statistics Container of
      <xref target="I-D.dikshit-grow-bmp-rd-scoped-rib-stats"/> with
      Family=EVPN, so the two documents share one wire format.  This
      changes the per-EVI Stat Data layout: it now carries Family, Subtype
      (the EVPN route type), AFI, SAFI, RD, and Gauge, replacing the
      earlier Stats Type / Stats Subtype / RD / Gauge layout.  The RD
      semantics, the VLAN-aware bundle handling, and the IRB counting rule
      are retained.</t>

	<t>Moved
      <xref target="I-D.dikshit-grow-bmp-rd-scoped-rib-stats"/> to
      Normative References, since the per-EVI encoding now depends on it.</t>

	<t>Updated the worked example (<xref target="sect-2.3"/>) to show the
      container inside the full Statistics Report nesting, and added
      <xref target="RFC9069"/> for the Loc-RIB Instance Peer context.</t>

	<t>Recast the per-RD multihoming and leaked-route text in
      <xref target="sect-2.2.1"/> as an open item, and noted that
      Ethernet Segment statistics are naturally scoped by ESI, a different
      axis not modeled here.</t>

	</list>
	</t>

	<t>
   v06 -&gt; v07</t>

	<t><list style="symbols"><t>Clarified in <xref target="sect-2.2.2"/> that the Route
      Distinguisher identifies the instance as seen in the monitored RIB:
      the per-EVI gauge is a per-EVI count on the Local-RIB (TBD_09) but a
      per-RD slice on the Adj-RIB-In and Adj-RIB-Out types (TBD_07,
      TBD_08, TBD_10, TBD_11), in particular on a Route Reflector or Route
      Server, consistent with
      <xref target="I-D.dikshit-grow-bmp-rd-scoped-rib-stats"/>.</t>

	<t>Documented, in <xref target="sect-2.2.2"/>, that a per-EVI gauge for
      an RFC 7432 VLAN-aware bundle service aggregates all broadcast
      domains sharing the RD, and noted an optional future Ethernet Tag ID
      selector.</t>

	<t>Specified, in <xref target="sect-2.2.2"/>, the counting rule for an
      EVPN IRB Type 2 route imported into both a MAC-VRF and an IP-VRF,
      and how a MAC-VRF and an IP-VRF that share one RD remain separable
      by Stats Subtype.</t>

	<t>Allowed, in <xref target="sect-2.2.1"/>, the leaked-route,
      multihoming, multihomed-EVI, and aliased-path statistics (Stats
      Types 1 through 4) to be reported per RD using the per-EVI format,
      not only at the global level.</t>

	<t>Added an Operational Considerations section
      (<xref target="sect-op"/>) covering reporting cardinality,
      rate-limiting, and the interpretation of counts exported from an
      Adj-RIB on a Route Reflector or Route Server after policy, Route
      Target Constrain <xref target="RFC4684"/>, and best-path or Add-Path
      selection.  Added <xref target="RFC4684"/> to Informative
      References.</t>

	</list>
	</t>

	</section>
	</back>

	</rfc>
	
