draft-ietf-grow-bmp-yang.xml

<?xml version="1.0" encoding="UTF-8"?>
<?rfc toc="yes"?>
<?rfc tocompact="yes"?>
<?rfc tocdepth="4"?>
<?rfc tocindent="yes"?>
<?rfc symrefs="yes"?>
<?rfc sortrefs="yes"?>
<?rfc comments="yes"?>
<?rfc inline="yes"?>
<?rfc compact="yes"?>
<?rfc subcompact="no"?>
<rfc category="std" ipr="trust200902" docName="draft-ietf-grow-bmp-yang-latest" submissionType='IETF'>
  <front>
    <title abbrev="BMP YANG Module">BMP YANG Module</title>
    <author fullname="Camilo Cardona" initials="C" surname="Cardona ">
      <organization>NTT</organization>
      <address>
        <postal>
          <street>164-168, Carrer de Numancia</street>

          <city>Barcelona</city>

          <code>08029</code>

          <country>Spain</country>
        </postal>

        <email>camilo@ntt.net</email>
      </address>
    </author>
    <author fullname="Paolo Lucente" initials="P." surname="Lucente">
      <organization>NTT</organization>
      <address>
        <postal>
          <street>Siriusdreef 70-72</street>
          <city>Hoofddorp</city>
          <code>2132</code>
          <country>Netherlands</country>
        </postal>
        <email>paolo@ntt.net</email>
      </address>
    </author>
    <author fullname="Thomas Graf" initials="T." surname="Graf">
      <organization>Swisscom</organization>
      <address>
        <postal>
          <street>Binzring 17</street>
          <city>Zurich 8045</city>
          <country>Switzerland</country>
        </postal>
        <email>thomas.graf@swisscom.com</email>
      </address>
    </author>     
    <author fullname="Benoit Claise" initials="B" surname="Claise">
      <organization>Huawei</organization>
      <address>
        <email>benoit.claise@huawei.com</email>
      </address>
    </author> 
    <date/>
    <area>OPS</area>
    <workgroup>GROW</workgroup>
    <abstract>
      <t> 
        This document proposes a YANG module for the configuration and monitoring of the BGP Monitoring Protocol (BMP).
      </t>
    </abstract>
  </front>

  <middle>
    <section title="Terminology" anchor="terminology">
       <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>    
       <t>
           Routing Information Bases, peers, monitoring stations, and
           initiation messages are defined in <xref target="RFC7854"/>.
      </t>    
    </section>   
    <section title="Model description" anchor="model_description">
       <t>
              This document specifies a YANG module for configuring and
           monitoring the BGP Monitoring Protocol (BMP)  <xref target="RFC7854"/>.  The model
           provides parameters for configuring the session to BMP monitoring stations; the
           selection of the BGP Routing Information Bases (RIBs) for Route Monitoring Messages; provides
           operational metrics and enables resetting of BMP monitoring sessions.
       </t> 

       <t>The model is included in <xref target="model-content"/>. In this
           section, we provide details and examples of each of its parts.</t>

       <t>The BMP yang model is placed at the root of the YANG tree.  At its
           upper level, the BMP model lists each monitoring station. Every
           monitoring station is identified by an ID, which is a string
           provided by the operator.</t>


    <section title="IP Connectivity" anchor="ip_connectivitu">
        <t> BMP allows for active and passive
            connections between the router and the BMP monitoring station as described in  <xref
                target="RFC7854" section="3.2" sectionFormat="of"/>. In
            an active connection, the router establishes the TCP connection to
            the monitoring station, while in a passive one, it is the monitoring
            station which initiates the connection. The BMP yang module
            provides options for both types of connection using a choice.</t>

        <t>We describe each type of connection option next, and provide
            examples of their configuration.</t>

    <section title="Active connection">

        <t>For an active connection, the IP address and port of the monitoring
            station, together with the local address MUST be provided. One can
            optionally provide the local port for establishing the connection.
            If the monitoring station is connected over a network-instance
            instead of the global one, this one must also be specified. An
            example of configuration is included in <xref target="active_connection_option"/>.</t>

        <figure  anchor="active_connection_option">
          <name>Active connection example</name>
          <sourcecode><![CDATA[
INSERT_TEXT_FROM_FILE(clean_examples/figure_active_connection_option.xml)
           ]]></sourcecode>
        </figure>

        <t>Note in the example from  <xref target="active_connection_option"/>
                that there is no network instance defined, so the connection is
                using the global network instance.</t>
    </section>
      <section title="Passive connection">
        <t>In a passive connection, the IP of the monitoring station, the local address and local port
            for the incoming connection must be specified. If the port of the
            monitoring station is provided, it MUST match the incoming
            connection. If the monitoring station is connected through a
            network-instance instead of the global one, this one MUST also be
            specified.</t>

        <t>An incoming connection not matching a valid entry MUST be ignored by
            the router.</t>

        <figure anchor="passive_connection_option">
          <name>Passive connection example</name>
          <sourcecode><![CDATA[
INSERT_TEXT_FROM_FILE(clean_examples/figure_passive_connection_option.xml)
           ]]></sourcecode>
          </figure>
      </section>
    </section>
    <section title="TCP Options" anchor="tcp_options">
        
        <t>The BMP module allows tuning various parameters of the TCP
            connection supporting the BMP session:</t>

      <t>
          <ul spacing="compact">
            <li>For configuring TCP keepalives, the connection container uses the tcp-common-grouping from <xref target="I-D.ietf-netconf-tcp-client-server"/>. Please see <xref target="I-D.ietf-netconf-tcp-client-server" section="2.1.3.1" sectionFormat="of"/> for the explanation of each of its parameters. Note that for the configuration of these parameters, the device must have the feature "tcp-client-keepalives" enabled. See also <xref target="RFC9293" section="3.8.4" sectionFormat="of"/></li>
            <li>The maximum segment of the TCP connection. See  <xref target="RFC9293" section="3.7.1" sectionFormat="of"/>.</li>
            <li>Enabling MTU discovery for the path. See <xref target="RFC9293" section="3.7.2" sectionFormat="of"/>.</li>
            <li>Session security. Provides options for authentication using AO and MD5. This part of the model was taken from the BGP yang model <xref target="I-D.ietf-idr-bgp-model"/>.</li>
        </ul>
      </t>

      <t>Figures <xref target="tcp_example1" format="counter"/> and <xref target="tcp_example2" format="counter"/> include examples configuring the previous TCP parameters in the model.</t>

        <t>
          <figure anchor="tcp_example1">
          <name>Example of configuring basic TCP parameters</name>
            <sourcecode><![CDATA[
INSERT_TEXT_FROM_FILE(clean_examples/figure_tcp_example1.xml)
           ]]></sourcecode>
          </figure>
        </t> 


        <t>
          <figure anchor="tcp_example2">
          <name>Example of the Configuration of TCP session security.</name>
            <sourcecode><![CDATA[
INSERT_TEXT_FROM_FILE(clean_examples/figure_tcp_example2.xml)
           ]]></sourcecode>
          </figure>
        </t> 

    </section>
    <section title="Other BMP connectivity options">

    <t>The model also includes  the next options to configure the connection to the BMP monitoring station:</t>

      <t>
        <ul spacing="compact">
            <li>Initial-delay: a value in seconds that the device must hold back before starting the connection to the station. </li>
            <li>Backoff time. Configuration of the backoff time strategy after failing to connect to the monitoring station. The model includes a basic exponential backoff with a default initial backoff of 30 seconds and a maximum of 720 seconds, as suggested in <xref target="RFC7854" section="3.2" sectionFormat="of"/>.</li>
        </ul>
      </t>

      <t>In the example in <xref target="example_indelay"/>, we configure an initial-delay of 10. Configuring an initial-backoff of 50 seconds and 600 of maximum-backoff.</t>

        <t>
          <figure anchor="example_indelay">
          <name>Example of the initial-delay and simple exponential backoff.</name>
            <sourcecode><![CDATA[
INSERT_TEXT_FROM_FILE(clean_examples/figure_example_indelay.xml)
           ]]></sourcecode>
          </figure>
        </t> 
    </section>
    <section title="BMP data">

        <t>The bmp-data container defines the configuration parameters for the data that the device sends to the monitoring station using the various BMP messages. See <xref target="RFC7854" section="4" sectionFormat="of"/>.</t>

        <t>The BMP model defines options for the initiation message, the statistics report, and the routing monitoring. The first two have simple configurations options and are shortly described next. The Routing monitoring is the most complex of all and it is detailed in <xref target='route_monitoring'/>.</t>

      <t>
        <ul spacing="compact">
            <li>Initiation-message: Content for an information TLV type-0 for identification of the device. See <xref target="RFC7854" section="4.3" sectionFormat="bare"/> and <xref target="RFC7854" section="4.4" sectionFormat="of"/></li> 
            <li>Statistics-interval: The statistics report is enabled by the presence of the bmp-statistics-report container. The statistics-interval is mandatory if the bmp-statistics-report container exists and defines the interval of the statistics report. See <xref target="RFC7854" section="4.8" sectionFormat="of"/>.</li>
        </ul>
      </t>

      <t>An example of configuring the previous options is included in <xref target="example_init_msg"/></t>

        <t>
          <figure anchor='example_init_msg'>
          <name>Example of configuration of initiation-message and statistics report interval.</name>
            <sourcecode><![CDATA[
INSERT_TEXT_FROM_FILE(clean_examples/figure_example_init_msg.xml)
           ]]></sourcecode>
          </figure>
        </t> 

        <section title="BMP route monitoring" anchor="route_monitoring">

        <t>Route monitoring messages are used for synchronization of RIBs to the monitoring station. See <xref target="RFC7854" section="5" sectionFormat="of"/>.</t>

        <t>The next 3 requirements were defined before designing this part of the model.</t>

      <t>
        <ul spacing="compact">
            <li>Operators might not want to receive all routes from all RIBs in a network device. For instance, some devices contain a considerable amount of data that might overwhelm the monitoring station. In this cases, operators might want to only collect information from an arbitrary subset of RIBs, address families, peers.</li>
            <li>Operators might want to configure the route monitoring messages for different network instances differently. For example, they might want to receive different address families from the global network instance than in L3 VPN network instances. </li>
            <li>In contrast to the previous points. some operators might want a simple configuration that covers multiple cases (e.g. same config for all peers, or same config for all network instances).  This would not only make configurations look smaller and concise, but will reduce the need for reconfiguring devices when you add a new peer or add a new network instance (which happens frequently on some type of networks). </li>
        </ul>
      </t>

      <t>Based on the previous points, the BMP yang model is designed to flexibly control the data sent through the BMP route monitoring packets, yet it provides options to facilitate configurations for simple cases, such as when the operator wants to receive all routes from a RIB.</t>

      <t>The Routing monitoring configuration is divided in a 4 part hierarchy:</t>

      <t>
        <ul spacing="compact">
            <li>Network Instance </li>
            <li>RIB Type (e.g. Adj-RIB-IN pre/post, local RIB)</li>
            <li>Address Family </li>
            <li>Peers</li>
        </ul>
      </t>

      <t>Absence of the routing monitoring container will disable the routing monitoring messages to the monitoring station.</t>

      <t>We'll offer an introduction to these hierarchies before going over them with detail.</t>

      <t>The number of RIB types (e.g. Adj-RIB-IN, etc/OUT and local RIB) and Address families is low, and their configuration should not change frequently. Therefore, they are configured explicitly in the model. That is, the model does not provide a way of providing a default configuration for these or configuring them in groups.</t>

      <t>On the other hand, Network instances and peers require greater flexibility.</t>

      <t>For network instances, the model should configure not only the "global" network instance (the main one configured under the /ietf-routing:routing), but also other network instances configured under the /ietf-network-instance:network-instances/. Also, network instances can change frequently in networks with customer connecting to Virtual Private Networks. To not force operators to change configuration at every change, the model provides methods for  defining a "default" configuration for network instances. However, to provide control over the configuration, each network instance can be configured independently, if needed.</t>

      <t>A situation is similar with peers for the Adj-RIB-IN and Adj-RIB-OUT RIBs. The model includes a way of configuring a default for all peers for simple cases, but one can provide configuration for type of peers, peergroups,  or each peer individually. </t>

      <t>We summarize the requirements stated on the previous two paragraphs next:</t>

      <t>For network instances:</t>

      <t>
        <ul spacing="compact">
          <li>The configuration should be simple for cases where only the "global" routing instance is enabled.</li>
          <li>The model should  provide ways of configuring all Network instances (kind of a default config for any Network instance that is configured in the device).</li>
          <li>The model should provide a way of configuring network instances individually.</li>
        </ul>
      </t>

      <t>For peers:</t>
      <t>
       <ul spacing="compact">
            <li>The model should provide ways of configuring all peers, kind of a default. This would be the most common case.</li>
            <li>The model should provide ways of configuring peergroups.</li>
            <li>The model should provide ways of configuring type of peers. For instance, only send routes from eBGP peers.</li>
            <li>The model should provide ways of configuring individual peers. For instance, turn a route-policy filtering prefix for a specific peer, or turning off a peer that is noisy yet not important.</li>
        </ul>
      </t>

      <t>To further control the route monitoring data, the peer/peer-type container includes a route-policy option in which the operator can further filter the data send to the BMP monitoring station.</t>

      <t>We'll describe each of the 4 hierarchies, and provide examples for each, in the next sections.</t>

      <section title="Network instances">

          <t>The routing monitoring configuration starts with the configuration of network instances. A network instance can be configured individual or it can be configured if it matches any of the selectors from the "bmp-ni-types" identity. We explain each next</t>

          <t>The model currently defines two bmp-ni-types identities: "bmp-ni-types-all-ni" which selects all network instances, and "bmp-ni-types-non-global-ni" which selects all network instances, except the global one. The former can be used as a "default" configuration for simple cases.</t>

          <t>The model offers two ways of configuring a network-instance individually. The main one is under container "bmp-data/bmp-monitoring-stations/" under the "routing/control-plane-protocols/control-plane-protocol/bgp". The second one is specifying the name under the "bmp/bmp-route-monitoring/network-instance". The former one is the prefered method, since it allows for better validation of the data. This method requires the support of schema-mount. The latter method does not validate some leafs, but it does not requires support for schema-mount by the device.</t>

          <t>An empty configuration disables routing monitoring messages for the selected network-instances. Operators can also use the "enable" leaf to disable explicitly the routing messages for the network instance. </t>

          <t>The route-monitoring data for a network instance can be configured by maximum one element of the network-instances list.  There SHOULD  be clear rules to which element to apply to a network instance in case multiple elements can select it. We provide rules and examples in the next part of the section.</t>

      <t>
        <ul spacing="compact">
            <li>If the BGP container under the network-instance includes a BMP/Route monitoring container, it SHOULD be configured by it. Note that if the monitoring-station is not present in the container, the route monitoring will be disabled for this network instance.</li>
            <li>If the bmp-ni-types-global-ni exist, the global network instance / network instance SHOULD be configured using this element.</li>
            <li>If any /ni:network-instances/ni:network-instance/ is referenced, the network instance SHOULD be configured using this element.</li>
            <li>Any Network instance not referenced by any rule above SHOULD be configured using the bmp-ni-types-all-ni if one exists. If it does not exist, then the network instance is not configured (and therefore no route monitoring messages from the network instance are sent to the monitoring station).</li>
        </ul>
      </t>

      <t>Any extension of the bmp-ni-types SHOULD provide explanations of how to deal with case in which multiple elements select the same network instance. </t>

      <t>We provide examples of configuring the network instance level next. For now, we will focus on the configuration using the BMP container (not the configuration under the BGP container).  To focus on the network instance configuration, we mask the configuration under each instance using "Configuration X". </t>


        <t>
          <figure anchor='example_netins_one'>
          <name>Examples of configuring the network instance level for Route Monitoring.</name>
            <sourcecode><![CDATA[
INSERT_TEXT_FROM_FILE(clean_examples/figure_example_netins_one.xml)
           ]]></sourcecode>
          </figure>
        </t> 

        <t>In example from <xref target="example_netins_one"/>, we have a "default" configuration (Configuration A) applied to any network instance without any explicit configuration. The global network instance and network-instance-two get Configuration B and Configuration C, respectively. The network-instance-one instance container disables the routing monitoring messages for that network instance.</t>


        <t>
          <figure anchor='simple_ni_configuration'>
          <name>Example of configuring all network instances.</name>
            <sourcecode><![CDATA[
INSERT_TEXT_FROM_FILE(clean_examples/figure_simple_ni_configuration.xml)
           ]]></sourcecode>
          </figure>
        </t> 

        <t>The example in <xref target="simple_ni_configuration"/> shows a "simple" configuration. In this case, all network instances would get "Configuration D". Note that `bmp-ni-types-all-ni` would also cover the global instance.</t>

        <t>Another simple configuration would just involve configuring the global network instance. In this case, information of non-global network instances would not be sent to the monitoring station. This is depicted in <xref target='simple_gni_configuration'/></t>

        <t>
          <figure  anchor='simple_gni_configuration'>
          <name>Example of configuring only the global network instance.</name>
            <sourcecode><![CDATA[
INSERT_TEXT_FROM_FILE(clean_examples/figure_simple_ni_configuration.xml)
           ]]></sourcecode>
          </figure>
        </t> 

    </section>
    <section title="RIB Type">

        <t>Alternatively to the configuration of <xref target='simple_gni_configuration'/>, one can configure directly the global instance, but using the BMP configuration under the BGP container</t>

    <t>Each RIB type is configured explicitly in the model through a container. The model currently provides containers for adj-rib-out-pre, adj-rib-out-post, adj-rib-in-post, adj-rib-in-pre and local-rib. </t>

    <t>An empty configuration or absence of a RIB-type container disables route-messages for it. Operators can also disable the RIB-type route monitoring messages by marking the "enabled" leaf as False.</t>

    <t>We provide an example of this, together with address families, in the next section</t>

    </section>
    <section title="Address families">

    <t>Address families are configured explicitly within each RIB-TYPE using a list. The key is of type `ietf-bgp-types:afi-safi-type` without any further constraint.</t>

    <t>An empty configuration or absence of an address family disables route-messages for it. Operators can also disable the address-family route monitoring messages by marking the "enabled" leaf as False.</t>

    <t>We show a few examples of configuring RIB-Types and Address families next. We will mask further configurations of address families with "Configuration X" to focus on the covered parts.</t>

        <t>
          <figure anchor='example_rib_af_one'>
          <name>Example of configuring RIBs and address families.</name>
            <sourcecode><![CDATA[
INSERT_TEXT_FROM_FILE(clean_examples/figure_example_rib_af_one.xml)
           ]]></sourcecode>
          </figure>
        </t> 

        <t>In <xref target='example_rib_af_one'/>,  we expand previous sections examples with RIB-Type and address families configurations. The expected result of the previous configuration would be:</t>

      <t>
        <ul spacing="compact">
            <li>For the global network instance, adj-rib-in-pre and adj-rib-in-post RIBs are enabled. In each of them IPv4 and IPv6 address families are configured. The configuration can be the same or not, depending on the requirements of the operators. Any other RIB and address families are disabled.</li>
            <li>Network instance one is disabled, meaning that routing monitoring messages are disabled for that network instance.</li>
            <li>Network instance "network-instance-two" has adj-rib-out-post enabled, but only address family ipv4-unicast is configured. The ipv6-unicast will not be configured for this instance.</li>
            <li>For all other network instances, adj-rib-in-pre with IPv4 and IPv6 address families are configured, thanks to the configuration of bmp-ni-types-all-ni</li>
        </ul>
      </t>


      <t>If an operator only wants to configure the IPv4/IPv6 of adj-rib-pre-in for the global instance,  the configuration in <xref target='example_rib_af_two'/> plus the peer configuration (coming in next section) will be enough. We note again that even if the configuration of both address families is the same, they must be explicitly configured for each of them.</t>

        <t>
        <figure anchor='example_rib_af_two'>
          <name>Example of configuring RIBs and address families.</name>
          <sourcecode><![CDATA[
INSERT_TEXT_FROM_FILE(clean_examples/figure_example_rib_af_two.xml)
           ]]></sourcecode>
          </figure>
        </t> 

    </section>
    <section title="Peers">

    <t>For adj-RIB-in and adj-RIB-out, both pre and post, the model requires the selection of peer's RIBs that will be transmitted to the monitoring station. The local-rib does not include this container.</t>

    <t>Peers are a list indexed by a peer id, which can be one of the following: </t>

      <t>
        <ul spacing="compact">
            <li>An individual peer, using a remote address. For the configuration under the "/bmp" tree, The model currently does not check if the remote address exists, that would be a responsibility of the device.</li>
            <li>Peergroups.</li>
            <li>A group of peers matching a BGP type. i.e. eBGP peers. </li>
            <li>One or more peers defined by an `bmp-peer-types` identity. The BMP model currently provides the `bmp-peer-types-all-peers` identity which select all peers. For simple cases, this is the value that would normally be considered.</li>
        </ul>
      </t>

    <t>Peers MUST be selected (configured) by maximum a single instance of the peers list. For the included keys in the BMP model, the process to select which instance to use is the next:</t>

      <t>
        <ul spacing="compact">
            <li>If there is a peer address matching the peer, it should be configured using that instance.</li>
            <li>If the peer is of any BGP type listed in the peer list, it should be configured using this instance.</li>
            <li>If there is a peer instance identified with the `bmp-peer-types-all-peers`, it would be configured using this instance.</li>
            <li>Finally, if no instance covers the peer, the data from this peer should not be transmitted to the monitoring station.</li>
        </ul>
      </t>

      <t>An empty configuration of a peer type disables route-messages for it. Operators can also disable the address-family route monitoring messages by marking the "enabled" leaf as False.</t>

      <t>Any additional bmp-peer-types identity created SHOULD describe how to unambiguously select a peer when there are conflicting options (multiple options covering the peer).</t>

      <t>We'll provide examples of the peers configuration after describing the filter containers.</t>

    </section>
    <section title="Filtering route-monitoring messages">

    <t>The local rib, and the peer containers within the rest of rib types, include a filter container. This container includes mechanisms to filter route-monitoring messages for the specific RIB.</t>

    <t>The policy-filter can include a routing policy that, if existing, should be applied to the outgoing updates to the monitoring station, and would serve as a granular way of filtering the messages that the monitoring station receives. </t>

    <t>Note that the policy-filter contains an `accept-route` default export policy. An operator can change it to a reject-route, if required.</t>

    <t>The policies created with the routing-policy can perform a large variety of actions routes, and can filter them based on multiple characteristics. For the consistency of the data in the monitoring station, the route policies actions SHOULD be restricted to accepting or rejecting routes. Furthermore, the conditions SHOULD only match prefix sets.</t>

    <t>We present examples of full configurations next.</t>
    </section>

    <section title="Full examples of Route monitoring configurations">


    <section title="Example one">

    <t>In the example configuration from  <xref target='full_example_one'/>, address families IPv6 and IPv4 are configured to send all peers. This is an example of a simple configuration</t>
        <t>
          <figure anchor="full_example_one">
              <name>Enabling Route monitoring for all peers; all network instances; IPv4/IPv6 Address families,  in the adj-rib-in-pre RIB.</name>
            <sourcecode><![CDATA[
INSERT_TEXT_FROM_FILE(clean_examples/figure_full_example_one.xml)
           ]]></sourcecode>
          </figure>
        </t> 

    </section>
    <section title="Example two">

        <t>In the example in <xref target='full_example_two'/>, the global network instance enables the adj-rib-in-pre. In this RIB, the IPv4 unicast address family is configured for all external peers. We assume peer 198.51.100.1 is external, but its BGP configuration is not shown in the snippet.  Peer 198.51.100.1, however, has a specific configuration: it announces everything but prefixes matching the test_policy list. Note that there is a default accept-route default policy in the model.</t>

        <t>
          <figure anchor="full_example_two">
            <name>Configuring address families differently for the global network instance</name>
            <sourcecode><![CDATA[
INSERT_TEXT_FROM_FILE(clean_examples/figure_full_example_two.xml)
           ]]></sourcecode>
          </figure>
        </t> 


    </section>
    <section title="Example three">

        <t>In the example from <xref target="full_example_three"/>, all network instances have adj-rib-in-pre with  IPv6  and IPv4  configured receiving all peers. network-instance-one is disabled, and network-instance-two is announcing only the local-rib/IPv4 unicast routes.</t>

        <t>
          <figure anchor="full_example_three">
            <name>Applying a general configuration to all network instances, except of two, which are configured specifically.</name>
            <sourcecode><![CDATA[
INSERT_TEXT_FROM_FILE(clean_examples/figure_full_example_three.xml)
           ]]></sourcecode>
          </figure>
        </t> 

    </section>


    </section> 
    </section> 
    </section> 
    <!-- End of BMP data-->

    <section title="Session stats">
        <t>The non-configurable container "session-stats" includes various metrics for the session with the monitoring station.</t>
    </section> 

    <section title="Session reset action">
        <t>The "session-reset" action resets a session with a monitoring station.</t>
    </section> 

    </section>  
    <!-- End of Model description -->


    <section title="Base ietf-bmp YANG module" anchor="model-content">
      <section title="Tree View" anchor="ietf-bmp-tree-view">
        <t>
        The following tree diagram provides an overview of the ietf-bmp.yang
            data model.
        </t>
        <t>
          <figure>
            <sourcecode><![CDATA[
INSERT_TEXT_FROM_FILE(ietf-bmp-trees.txt)
           ]]></sourcecode>
          </figure>
        </t> 
      </section>    

      <section title="YANG Module" anchor="ietf-bmp">
        <t>&lt;CODE BEGINS> file "ietf-bmp@2022-01-27.yang"</t>
          <figure>
            <sourcecode><![CDATA[
INSERT_TEXT_FROM_FILE(ietf-bmp.yang)
            ]]></sourcecode>
          </figure>
        <t>&lt;CODE ENDS></t>
      </section> 
    </section>  
                    
    <section title="Security Considerations" anchor="security-considerations">
      <t> 
         The YANG module specified in this document defines a schema for data
          that is designed to be accessed via network management protocols such
          as NETCONF <xref target="RFC6241"/> or RESTCONF <xref
          target="RFC8040"/>. The lowest NETCONF layer is the secure transport
          layer, and the mandatory-to-implement secure transport is Secure
          Shell (SSH) <xref target="RFC6242"/>. The lowest RESTCONF layer is
          HTTPS, and the mandatory-to-implement secure transport is TLS <xref
          target="RFC8446"/>. The NETCONF Access Control Model (NACM) <xref
              target="RFC8341"/> provides the means to restrict access for
          NETCONF or RESTCONF users to a preconfigured subset of all
          available NETCONF or RESTCONF protocol operations and content. 
      </t>
      <t>
         BGP data is sensible for security considerations. The model described
          in this document could be used to send BGP information to malicious
          BMP stations. Write access to this model SHOULD therefore be
          properly protected. 
      </t>
      <t>
          The session-reset action can demand considerable amount of resources
          from network elements. It SHOULD thus be protected from illegal access.
      </t>
    </section>

    <section title="IANA Considerations" anchor="iana-considerations">
      <section title="The IETF XML Registry">
      	<t>
      	   This document registers a URIs in the IETF XML 
      	   registry <xref target="RFC3688"/>.  Following the format in 
           <xref target="RFC3688"/>, the following registrations are
           requested:</t>
        <t>
          <figure>
          	<sourcecode>
   URI: urn:ietf:params:xml:ns:yang:ietf-bmp
   Registrant Contact: The IESG.
   XML: N/A, the requested URI is an XML namespace.
             </sourcecode>
           </figure>
         </t>
      </section>
      <section title="The YANG Module Name Registration">
        <t>
           This document registers the following YANG module in the "
           YANG Module Names" registry  
           registry <xref target="RFC6020"/>:</t>
        <t>
          <figure>
            <sourcecode>
   Name: ietf-bmp
   Namespace: urn:ietf:params:xml:ns:yang:ietf-bmp
   Prefix: bmp
   Reference: [This RFC-to-be]
             </sourcecode>
           </figure>
         </t>
      </section>
    </section>

    <section title="Open Issues">
      <t>
        <list counter="a">
          <t>The security considerations section will have to be aligned with 
              https://trac.ietf.org/trac/ops/wiki/yang-security-guidelines</t>
        </list>
      </t>
    </section>
  </middle>

  <back>
    <references title="Normative References"> 
      <?rfc include="reference.RFC.2119.xml"?>
      <?rfc include="reference.RFC.3688.xml"?>
      <?rfc include="reference.RFC.6241.xml"?>
      <?rfc include="reference.RFC.6242.xml"?>
      <?rfc include="reference.RFC.6991.xml"?>
      <?rfc include="reference.RFC.7854.xml"?>
      <?rfc include="reference.RFC.8671.xml"?>
      <?rfc include="reference.RFC.9069.xml"?>
      <?rfc include="reference.RFC.8040.xml"?>
      <?rfc include="reference.RFC.8174.xml"?>
      <?rfc include="reference.RFC.8349.xml"?>
      <?rfc include="reference.RFC.8446.xml"?>
      <?rfc include="reference.RFC.8341.xml"?>
      <?rfc include="reference.RFC.6020.xml"?>
      <?rfc include="reference.RFC.8529.xml"?>
      <?rfc include="reference.RFC.1191.xml"?>
      <?rfc include="reference.RFC.8177.xml"?>
      <?rfc include="reference.RFC.9293.xml"?>
      <?rfc include='reference.I-D.ietf-idr-bgp-model.xml'?>
      <?rfc include='reference.I-D.ietf-netconf-tcp-client-server.xml'?>
    </references>
<!--
    <references title="Informative References">
    </references>
        -->
    <?rfc needLines="100"?>

<!--   
    <section title="Changes between revisions">
      </t>   
      <t>v00 - v01
        <list style="symbols">
          <t>Placeholder: xxx</t>
          <t>Placeholder: yyy</t>
        </list>
      </t>
    </section>    
  -->

    <section title="Examples">
        <t>This sections shows some examples of BMP configuration using the
            model.</t>
        <section title="Example one">
            <t>In this example, the device connects to a monitoring station
            using an active connection. 
            The devices sends route monitoring messages for the global
            instance, the adj-rib-out-pre RIB, the IPv4/IPv6 address family,
            and external peers.
            </t>
        <t><figure anchor="example_one">
            <sourcecode><![CDATA[
INSERT_TEXT_FROM_FILE(clean_examples/figure_example_one.xml)
]]></sourcecode>         
    </figure></t>
    </section>
        <section title="Example two">
            <t>In the next example, the device connects to a monitoring station
            using a passive connection, over the network-instance monitoring.
            The configuration of route monitoring messages is more complex than
            in the previous example.  It shows how to combine the configuration
            of general identities of network instances and peers (e.g.
            bmp-ni-types-all-ni for NI, external for peers), and individual
            configurations to support a more complex requirement. This is what
                the example expects to configure:</t>
        <t>
        <list style="symbols">
            <t> For the global network instance,
            the device sends updates for adj-rib-in-pre, address families IPv4
            and IPv6. It sends updates for all external peers except peer
            128.66.1.1, which is disabled.</t>
        <t>Network instance monitoring is disabled for route monitoring messages. </t>
        <t>For the rest of network
            instances, we are enabling messages from adj-rib-in-pre, address
                families IPv4/IPv6, and for all peers.</t>
        </list>
        </t>
        <t><figure anchor="example_two">
            <sourcecode><![CDATA[
INSERT_TEXT_FROM_FILE(clean_examples/figure_example_two.xml)
]]></sourcecode>         
    </figure></t>
    </section>
    </section>

    <section title="Acknowledgements" numbered="no">
        <t>
          The authors would like to thank Yimin Shen, Jeff Haas, Pierre Vander
            Vorst, and Tom Petch for their review and feedback.
        </t>
    </section>

  </back>
</rfc>
<!-- Local Variables: -->
<!-- fill-column:72 -->
<!-- End: -->