Test Intent:
============

   Ref: https://github.com/ietf-wg-idr/draft-ietf-idr-bgp-ct/issues/14
        https://github.com/ietf-wg-idr/draft-ietf-idr-bgp-ct/issues/2

   BGP-CT carries transport class as an attribute, which impacts BGP update packing.
   Measure the impact of reduced Update packing on BGP control-plane convergence for
   scale conditions mentioned in:

   https://datatracker.ietf.org/doc/html/draft-hr-spring-intentaware-routing-using-color-00#section-6.3.2

   viz:

     "Support a massive scaled transport network
        *  Number of Remote PE's: >= 300k
        *  Number of Colors C: >= 5"

  Note: There are other issues related to MPLS label allocation and forwarding scale
    for this scale. This experiment does not focus on those. It focuses on only the 
    BGP control plane convergence aspect of this scaled scenario.

Environment:
=============
    All nodes are simulated as Docker containers on Linux server with 56 CPUs
    (server not dedicated to this experiment, has other workloads too).

    root@rahi-03:~# uname -nar
    Linux rahi-03 4.15.0-159-generic #167-Ubuntu SMP Tue Sep 21 08:55:05 UTC 2021 x86_64 x86_64 x86_64 GNU/Linux

    root@rahi-03:~# docker --version
    Docker version 20.10.7, build 20.10.7-0ubuntu5~18.04.3

    root@rahi-03:~# clab version | grep version
    version: 0.32.3

    Junos Routing Image: crpd:22.4R1.5

Topology:
=========

                           RR1 6
                            | 
                     5    4 |+------PE1 1
                    PE0-----P1------PE2 2
                             +------PE3 3


         - PE0-PE3 have 5 transport classes configured. All peer with RR1
         - PE1-PE3 are originating BGP-CT routes for 387105 endpoints in 5 transport classes.
         - RR1 reflects the 1.9M routes to PE0

Route Mix:
===========

  Transport route mix is used, each PE originates BGP CT transport routes for
  endpoints (e.g. loopback) that are used as nexthop in BGP service route
  updates. These BGP CT routes belong to 5 transport classes, hence each route
  is advertised with one of the 5 transport-target extended communities, as
  describe above, which impacts packing. Routes with same transport-target may
  be packed together. Routes with dissimilar transport-target will not be be
  packed in same Update.

Test Methodology:
=================

    Few Iterations of:
	 - Deactivate BGP at PE0. This brings down session with RR1.
         - Wait for cleanup complete.
         - Activate BGP at PE0. Wait for receiving all 1.9M routes from RR1.
         - Record the time taken to receive EOR, in-Updates, in-Octets
           shown in show bgp neighbor extensive output (PS1).


    Test independently for
        - BGP-CTv4 family,
        - BGP-CTv6 family,
        - BGP-CTv6 family with SRv6-SIDs.

 PS1: Example output:

  root@PE0# run show bgp neighbor extensive | match "family|table|update|initial|prefix"
  Nov 23 20:46:51
    Options: <LocalAddress AddressFamily Rib-group Refresh>
    Table bgp.transport-inet6.3 Bit: 20000
      Active prefixes:              1935480
      Received prefixes:            1935480
      Accepted prefixes:            1935480
      Advertised prefixes:          0
    Family inet6-transport
      Initial send time: 0
      Initial recv time: 58
    Input messages:  Total 13833  Updates 13831   Refreshes 0     Octets 55507402
    Output messages: Total 4      Updates 0       Refreshes 0     Octets 87


TEST RESULTS:
---------------

Image: crpd:23.1I20221101_2159_kaliraj
========================================

  BGP CT IPv4.

  Iteration       Num-TCs        Num-Endpoints    Total-Routes     In-Updates     Time
  -----------+----------------+----------------+---------------+---------------+-------
    1               5              387105          1935480          7851          73 sec
    2               5              387105          1935480          7851          82 sec
    3               5              387105          1935480          7851          77 sec
    4               5              387105          1935480          7851          76 sec
    5               5              387105          1935480          7851          75 sec


  BGP CT IPv6.

  Iteration       Num-TCs        Num-Endpoints    Total-Routes     In-Updates     Time
  -----------+----------------+----------------+---------------+---------------+-------
    1               5              387105          1935480          13831         73 sec
    2               5              387105          1935480          13831         65 sec
    3               5              387105          1935480          13831         66 sec
    4               5              387105          1935480          13832         71 sec
    5               5              387105          1935480          13832         64 sec
    6               5              387105          1935480          13832         64 sec


Junos Routing Image: crpd:22.4R1.5
========================================

  BGP CT IPv4.

   Iter     Num-TCs   Num-Endpoints    Total-Routes     In-Updates    In-Octets   Time
  -------+---------+----------------+---------------+---------------+-----------+-------
    1         5         387105          1935480           7851         31619342   72 sec
    2         5         387105          1935480           7851         31619342   94 sec
    3         5         387105          1935480           7851         31619342   72 sec
    4         5         387105          1935480           7851         31619342   72 sec
    5         5         387105          1935480           7851         31619342   72 sec
    6         5         387105          1935480           7851         31619342   71 sec (tcp-nodelay)
    7         5         387105          1935480           7851         31619342   72 sec (tcp-nodelay)
    8         5         387105          1935480           7851         31619342   94 sec (tcp-nodelay)



   BGP CT IPv6.

   Iter     Num-TCs   Num-Endpoints    Total-Routes     In-Updates    In-Octets   Time
  -------+---------+----------------+---------------+---------------+-----------+-------
    1         5         387105          1935480          13831         55507402    58 sec
    2         5         387105          1935480          13831         55507402    59 sec
    3         5         387105          1935480          13831         55507402    58 sec
    4         5         387105          1935480          13832         55507497    57 sec
    5         5         387105          1935480          13832         55507497    64 sec
    6         5         387105          1935480          13832         55507497    63 sec (tcp-nodelay)
    7         5         387105          1935480          13832         55507497    63 sec (tcp-nodelay)
    8         5         387105          1935480          13831         55507402    62 sec (tcp-nodelay)

 Document History
 =================

 ----------+------------+-------------------------+----------------------------
  Version       Date             Author                Comment
 ----------+------------+-------------------------+----------------------------
   1.0       12/22/2022   Kaliraj Vairavakkalai     First cut
 ----------+------------+-------------------------+----------------------------
   1.1       06/25/2023   Kaliraj Vairavakkalai     Sue Hares review comment
 ----------+------------+-------------------------+----------------------------