multicast_topology.tex

xorp源码hg

%
% $XORP: xorp/docs/user_manual/multicast_topology.tex,v 1.6 2006/03/09 06:42:10 pavlin Exp $
%

\chapter{Multicast Topology Discovery}

\section{Terminology and Concepts}

Multicast routing protocols such as PIM-SM (Protocol Independent
Multicast Sparse-Mode) and PIM-DM (Protocol Independent Multicast
Dense-Mode) build the multicast delivery tree by using the RPF
(Reverse-Path Forwarding) information toward the root of the tree.
The root could be the so-called Rendezvous Point (RP) (in case of
PIM-SM) or the source itself (in case of PIM-SM or PIM-DM).

The RPF information in each router is per multicast distribution tree
and is basically the next-hop neighbor router information toward the
root of the tree. In other words, the RPF router is the next-hop router
toward the root. In case of PIM-SM, the RPF neighbor is typically the
router that a Join message is sent to.

Obviously, all multicast routers must have consistent RPF state,
otherwise a Join message may never reach the root of the tree.
Typically, the unicast path forwarding information is used to create the
RPF information, because under normal circumstances the unicast routing
provides the necessary information to all routers.

Note that he unicast-based RPF creates multicast distribution trees
where each branch of the tree follows the unicast path from each leaf of
the tree toward the root. Usually this is the desired behavior, but
occasionally someone may want the unicast and the multicast traffic to
use different paths. For example, if a site has two links to its network
provider, one of the links may be used for unicast only, and the other
one only for multicast.

To provide for such flexibility in the configuration, the PIM-SM and
PIM-DM specifications use the so called Multicast Routing Information
Base (MRIB) for obtaining the RPF information.
Typically, the MRIB may be derived from the unicast routing table, but
some protocols such as MBGP may carry multicast-specific topology
information. Furthermore, the MRIB may be modified locally in each site
by taking into account local configuration and preferences.
A secondary function of the MRIB is to provide routing metrics for
destination addresses. Those metrics are used by the PIM-SM and PIM-DM
Assert mechanism.

\section{Configuring the MRIB}

The XORP RIB module contains a table with the MRIB. That table is
propagated to the PIM-SM module and is used by PIM-SM in the RPF
computation. The MRIB table inside the RIB module is completely
independent from the Unicast Routing Information Base (URIB) table.
The URIB table is created from the unicast routes calculated by unicast
routing protocols such as BGP, OSPF and RIP.
The MRIB table is created similarly, but only by those protocols that
are explicitly configured to add their routes to the MRIB. For example,
if Multi-protocol BGP is enabled, then the BGP module will add
multicast-specific routes to the MRIB.

Currently, XORP supports the following methods for adding routing
entries to the MRIB:

\begin{itemize}

  \item {\tt Multi-protocol BGP:} The BGP module can be configured to
  negotiate multiprotocol support with its peers. Then, the BGP
  multicast routes will be installed in the MRIB. See
  Chapter~\ref{bgp} for information how to configure BGP.

  \item {\tt Static Routes:} The Static Routes module can be used to
  configure static unicast and multicast routes. The unicast routes are
  added to the Unicast RIB, while the multicast routes are added to the
  MRIB. See Chapter~\ref{static_routes} for information how to configure
  Static Routes.

  \item {\tt FIB2MRIB:} If there are no unicast routing protocols
  configured in XORP to supply the MRIB routes, then the FIB2MRIB module
  can be used to populate the MRIB. If the FIB2MRIB module is enabled,
  it will register with the FEA to read the whole unicast forwarding
  table from the underlying system, and to receive notifications for all
  future modifications of that table. In other words, the FIB2MRIB task
  is to replicate the unicast forwarding information on that router into
  the MRIB. The FIB2MRIB module can be enabled by the following
  configuration statements:~\footnote{Note that prior to XORP
  Release-1.1, the {\tt enable} flag was used instead of {\tt disable}
  to enable or disable each part of the configuration.}

\vspace{0.1in}
%\noindent\framebox[\textwidth][l]{\scriptsize
\noindent\framebox[6in][l]{\scriptsize
\begin{minipage}{6in}
\begin{alltt}
\begin{tabbing}
xx\=xx\=xx\=xx\=xx\=\kill
protocols \{\\
\>fib2mrib \{\\
\>\>disable: false\\
\>\}\\
\}
\end{tabbing}
\end{alltt}
\end{minipage}
}
\vspace{0.1in}

\end{itemize}


\section{Monitoring the MRIB}

All operational commands for monitoring MRIB begin with
{\tt show route table}. This section describes those commands in details.

All RIB commands to view the RIB's inner tables have the following form:

{\stt show route table \{ipv4 | ipv6\} \{unicast | multicast\} <protocol>}

The commands to view the MRIB have the following form:

{\stt show route table \{ipv4 | ipv6\} multicast <protocol>}

The value of {\stt <protocol>} has to be one of the following:

\begin{itemize}

  \item {\tt ebgp} to show eBGP MBGP routes.

  \item {\tt fib2mrib} to show unicast routes for multicast extracted
  from kernel.

  \item {\tt final} to show MRIB winning routes.

  \item {\tt ibgp} to show iBGP MBGP routes.

  \item {\tt static} to show MRIB static routes.

\end{itemize}

For example, the following command can be used to show the IPv4 FIB2MRIB
routes:

\vspace{0.1in}
\noindent\framebox[\textwidth][l]{\scriptsize
\begin{minipage}{6in}
\begin{alltt}
\begin{tabbing}
xxxxxxxxxxxxxxxx\=\kill
user@hostname> \textbf{show route table ipv4 multicast fib2mrib}\\
10.2.0.0/24    \>[fib2mrib(254)/65535]\\
               \>> to 0.0.0.0 via dc2/dc2\\
10.3.0.0/24    \>[fib2mrib(254)/65535]\\
               \>> to 0.0.0.0 via dc1/dc1\\
10.4.0.0/24    \>[fib2mrib(254)/65535]\\
               \>> to 10.3.0.2 via dc1/dc1\\
10.5.0.0/24    \>[fib2mrib(254)/65535]\\
               \>> to 10.2.0.4 via dc2/dc2\\
10.6.0.0/24    \>[fib2mrib(254)/65535]\\
               \>> to 10.2.0.1 via dc2/dc2
\end{tabbing}
\end{alltt}
\end{minipage}
}
\vspace{0.1in}

The final MRIB table can be shown with the following command:

\vspace{0.1in}
\noindent\framebox[\textwidth][l]{\scriptsize
\begin{minipage}{6in}
\begin{alltt}
\begin{tabbing}
xxxxxxxxxxxxxxxx\=\kill
user@hostname> \textbf{show route table ipv4 multicast final}\\
10.2.0.0/24    \>[connected(0)/0]\\
               \>> to 0.0.0.0 via dc2/dc2\\
10.3.0.0/24    \>[connected(0)/0]\\
               \>> to 0.0.0.0 via dc1/dc1\\
10.4.0.0/24    \>[fib2mrib(254)/65535]\\
               \>> to 10.3.0.2 via dc1/dc1\\
10.5.0.0/24    \>[fib2mrib(254)/65535]\\
               \>> to 10.2.0.4 via dc2/dc2\\
10.6.0.0/24    \>[fib2mrib(254)/65535]\\
               \>> to 10.2.0.1 via dc2/dc2
\end{tabbing}
\end{alltt}
\end{minipage}
}
\vspace{0.1in}