Merge commit '78986c0' into tmp-3.0-master-merge

[mirror_frr.git] / doc / vnc.texi

@c -*-texinfo-*-
@c This is part of the Frr Manual.
@c @value{COPYRIGHT_STR}
@c See file frr.texi for copying conditions.

@node VNC and VNC-GW
@chapter VNC and VNC-GW
This chapter describes how to use
Virtual Network Control (@acronym{VNC}) services,
including Network Virtualization Authority (@acronym{NVA}) and 
VNC Gateway (@acronym{VNC-GW}) functions.
Background information on NVAs, 
Network Virtualization Edges (@acronym{NVE}s), underlay networks (@acronym{UN}s),
and virtual networks (@acronym{VN}s) is available from the  
@url{https://datatracker.ietf.org/wg/nvo3,IETF Network Virtualization Overlays (@acronym{NVO3}) Working Group}.
VNC Gateways (@acronym{VNC-GW}s) support the import/export of routing
information between VNC and customer edge routers (@acronym{CE}s)
operating within a VN.  Both IP/Layer 3 (L3) VNs, and IP with
Ethernet/Layer 2 (L2) VNs are supported.

BGP, with IP VPNs and Tunnel Encapsulation, is used to distribute VN
information between NVAs. BGP based IP VPN support is defined in
@cite{RFC4364, BGP/MPLS IP Virtual Private Networks (VPNs)}, and
@cite{RFC4659, BGP-MPLS IP Virtual Private Network (VPN) Extension for
IPv6 VPN }.  Both the Encapsulation Subsequent Address Family Identifier
(SAFI) and the Tunnel Encapsulation Attribute, @cite{RFC5512, The BGP
Encapsulation Subsequent Address Family Identifier (SAFI) and the BGP
Tunnel Encapsulation Attribute}, are supported.

The protocol that is used to communicate routing and Ethernet / Layer 2
(L2) forwarding information between NVAs and NVEs is referred to as the
Remote Forwarder Protocol (RFP). @code{OpenFlow} is an example
RFP.  Specific RFP implementations may choose to implement either a
@code{hard-state} or @code{soft-state} prefix and address registration
model.  To support a @code{soft-state} refresh model, a @var{lifetime}
in seconds is associated with all registrations and responses.

The chapter also provides sample configurations for basic example scenarios.  

@menu
* Configuring VNC Services::
* Manual Address Control::
* Other VNC-Related Commands::
* Example VNC and VNC-GW Configurations::
* Release Notes::
@end menu

@node Configuring VNC Services
@section Configuring VNC

Virtual Network Control (@acronym{VNC}) service configuration commands
appear in the @code{router bgp} section of the BGPD configuration file
(@pxref{BGP Configuration Examples}). The commands are broken down into
the following areas:

@menu
* General VNC Configuration::
* RFP Related Configuration::
* VNC Defaults Configuration::
* VNC NVE Group Configuration::
* VNC L2 Group Configuration::
* Configuring Redistribution of Routes from Other Routing Protocols::
* Configuring Export of Routes to Other Routing Protocols::
@end menu

@code{General VNC} configuration applies to general VNC operation and is
primarily used to control the method used to advertise tunnel
information.  

@code{Remote Forwarder Protocol (RFP)} configuration relates to the
protocol used between NVAs and NVEs.  

@code{VNC Defaults} provides default parameters for registered NVEs.

@code{VNC NVE Group} provides for configuration of a specific set of 
registered NVEs and overrides default parameters.

@code{Redistribution} and @code{Export} control VNC-GW operation, i.e.,
the  import/export of routing
information between VNC and customer edge routers (@acronym{CE}s)
operating within a VN.

@node General VNC Configuration
@subsection General VNC Configuration

@deffn {VNC} {vnc advertise-un-method encap-safi|encap-attr} {}
Advertise NVE underlay-network IP addresses using the encapsulation SAFI
(@code{encap-safi}) or the UN address sub-TLV of the Tunnel Encapsulation attribute
(@code{encap-attr}). When @code{encap-safi} is used, neighbors under 
@code{address-family encap} and/or @code{address-family encapv6} must be
configured.  The default is @code{encap-attr}. 
@end deffn

@node RFP Related Configuration
@subsection RFP Related Configuration 

The protocol that is used to communicate routing and Ethernet / L2
forwarding information between NVAs and NVEs is referred to as the
Remote Forwarder Protocol (RFP).  Currently, only a simple example RFP
is included in Frr.  Developers may use this example as a starting
point to integrate Frr with an RFP of their choosing, e.g.,
@code{OpenFlow}.  The example code includes the following sample
configuration: 

@deffn {RFP} {rfp example-config-value @var{VALUE}} 
This is a simple example configuration parameter included as part of the
RFP example code.  @code{VALUE} must be in the range of 0 to 4294967295.
@end deffn

@node VNC Defaults Configuration
@subsection VNC Defaults Configuration

The VNC Defaults section allows the user to specify default values for
configuration parameters for all registered NVEs.
Default values are overridden by @ref{VNC NVE Group Configuration}. 

@deffn {VNC} {vnc defaults} {}
Enter VNC configuration mode for specifying VNC default behaviors.  Use
@code{exit-vnc} to leave VNC configuration mode.  @code{vnc
defaults} is optional.

@example
vnc defaults
  ... various VNC defaults
exit-vnc
@end example
@end deffn

These are the statements that can appear between @code{vnc defaults}
and @code{exit-vnc}.

@deffn {VNC} {rt import @var{rt-list}} {}
@deffnx {VNC} {rt export @var{rt-list}} {}
@deffnx {VNC} {rt both @var{rt-list}} {}

Specify default route target import and export lists.  @var{rt-list} is a
space-separated list of route targets, each element of which is
in one of the following forms:
@itemize
@item @var{IPv4-address}:@var{two-byte-integer}
@item @var{four-byte-autonomous-system-number}:@var{two-byte-integer}
@item @var{two-byte-autonomous-system-number}:@var{four-byte-integer}
@end itemize

If no default import RT list is specified, then the default import RT
list is empty.
If no default export RT list is specified, then the default export RT
list is empty.

A complete definition of these parameters is
given below (@pxref{VNC NVE Group Configuration}).

@end deffn

@deffn {VNC} {rd @var{route-distinguisher}}

Specify the default route distinguisher (RD) for routes advertised via BGP
VPNs.  The route distinguisher must be in one of four forms:
@itemize
@item @var{IPv4-address}:@var{two-byte-integer}
@item @var{four-byte-autonomous-system-number}:@var{two-byte-integer}
@item @var{two-byte-autonomous-system-number}:@var{four-byte-integer}
@item auto:vn:@var{two-byte-integer}
@end itemize

If RD is specified in the defaults section, the default RD
value is @var{two-byte-autonomous-system-number=0}:@var{four-byte-integer=0}.

A complete definition of this parameter is
given below (@pxref{VNC NVE Group Configuration}).
@end deffn

@deffn {VNC} {l2rd @var{nve-id-value}}
Set the value used to distinguish NVEs connected to the same logical
Ethernet segment (i.e., L2VPN).

A complete definition of this parameter is
given below (@pxref{VNC NVE Group Configuration}).
@end deffn

@deffn {VNC} {response-lifetime @var{lifetime}|infinite} {}
Specify the default lifetime to be included in RFP
response messages sent to NVEs.

A complete definition of this parameter is
given below (@pxref{VNC NVE Group Configuration}).

@end deffn

@deffn {VNC} {export bgp|zebra route-map MAP-NAME}
Specify that the named route-map should be applied to routes
being exported to bgp or zebra.
@end deffn

@deffn {VNC} {export bgp|zebra no route-map}
Specify that no route-map should be applied to routes
being exported to bgp or zebra.
@end deffn

@deffn {VNC} {export bgp|zebra ipv4|ipv6 prefix-list LIST-NAME}
Specify that the named prefix-list filter should be applied to
routes being exported to bgp or zebra.
Prefix-lists for ipv4 and ipv6 are independent of each other.
@end deffn

@deffn {VNC} {export bgp|zebra no ipv4|ipv6 prefix-list}
Specify that no prefix-list filter should be applied to
routes being exported to bgp or zebra.
@end deffn

@deffn {VNC} {exit-vnc} {}
Exit VNC configuration mode.
@end deffn

@c The following example @code{vnc defaults} defines a route target import-export
@c list for the route targets 1000:1 and 1000:2; a default route
@c distinguisher, 4444:10; and a default response lifetime of 500
@c seconds.
@c 
@c @example
@c vnc defaults
@c     rt both 1000:1 1000:2
@c     rd 4444:10
@c     response-lifetime 500
@c exit-vnc
@c @end example

@node VNC NVE Group Configuration
@subsection VNC NVE Group Configuration

A NVE Group corresponds to a specific set of NVEs.  A Client NVE is
assigned to an NVE Group based on whether there is a match for either
its virtual or underlay network address against the VN and/or UN address
prefixes specified in the NVE Group definition.  When an NVE Group
definition specifies both VN and UN address prefixes, then an NVE must
match both prefixes in order to be assigned to the NVE Group.  In the
event that multiple NVE Groups match based on VN and/or UN addresses,
the NVE is assigned to the first NVE Group listed in the configuration.  
If an NVE is not assigned to an NVE Group, its messages will be ignored.

Configuration values specified for an NVE group apply to all
member NVEs and override configuration values specified in the VNC
Defaults section.

@strong{At least one @code{nve-group} is mandatory for useful VNC
operation.}

@deffn {VNC} {vnc nve-group @var{name}} {}
Enter VNC configuration mode for defining the NVE group @var{name}.  
Use @code{exit} or @code{exit-vnc} to exit group configuration mode.

@example
vnc nve-group group1
  ... configuration commands
exit-vnc
@end example
@end deffn

@deffn {VNC} {no vnc nve-group @var{name}} {}
Delete the NVE group named @var{name}.
@end deffn

The following statements are valid in an NVE group definition:

@deffn {VNC} {l2rd @var{nve-id-value}}
Set the value used to distinguish NVEs connected to the same physical
Ethernet segment (i.e., at the same location)@footnote{The nve-id is
carried in the route
distinguisher.  It is the second octet of the eight-octet route
distinguisher generated for Ethernet / L2 advertisements.
The first octet is a constant 0xFF, and the third through eighth
octets are set to the L2 ethernet address being advertised.}

The nve-id subfield may be specified as either a literal value
in the range 1-255, or it may be specified as @code{auto:vn}, which
means to use the least-significant octet of the originating
NVE's VN address.
@end deffn

@deffn {VNC} {prefix vn|un A.B.C.D/M|X:X::X:X/M} {}
@anchor{prefix}
Specify the matching prefix for this NVE group by either virtual-network address
(@code{vn}) or underlay-network address (@code{un}). Either or both virtual-network
and underlay-network prefixes may be specified.  Subsequent virtual-network or
underlay-network values within a @code{vnc nve-group} @code{exit-vnc}
block override their respective previous values.

These prefixes are used only for determining assignments of NVEs
to NVE Groups.
@end deffn

@deffn {VNC} {rd @var{route-distinguisher}}
Specify the route distinguisher for routes advertised via BGP
VPNs.  The route distinguisher must be in one of these forms:
@itemize
@item @var{IPv4-address}:@var{two-byte-integer}
@item @var{four-byte-autonomous-system-number}:@var{two-byte-integer}
@item @var{two-byte-autonomous-system-number}:@var{four-byte-integer}
@item auto:vn:@var{two-byte-integer}
@end itemize

Routes originated by NVEs in the NVE group will use
the group's specified @var{route-distinguisher} when they are
advertised via BGP. 
If the @code{auto} form is specified, it means that a matching NVE has
its RD set to
@var{rd_type=IP=1}:@var{IPv4-address=VN-address}:@var{two-byte-integer},
for IPv4 VN addresses and
@var{rd_type=IP=1}:@var{IPv4-address=Last-four-bytes-of-VN-address}:@var{two-byte-integer},
for IPv6 VN addresses.

If the NVE group definition does not specify a @var{route-distinguisher},
then the default @var{route-distinguisher} is used.
If neither a group nor a default @var{route-distinguisher} is
configured, then the advertised RD is set to
@var{two-byte-autonomous-system-number=0}:@var{four-byte-integer=0}.
@end deffn

@deffn {VNC} {response-lifetime @var{lifetime}|infinite} {}
Specify the response lifetime, in seconds, to be included in RFP
response messages sent to NVEs.  If the value
``infinite'' is given, an infinite lifetime will be used.

Note that this parameter is not the same as the lifetime supplied by
NVEs in RFP registration messages. This parameter does not affect
the lifetime value attached to routes sent by this server via BGP.

If the NVE group definition does not specify a @var{response-lifetime},
the default @var{response-lifetime} will be used.
If neither a group nor a default @var{response-lifetime} is configured,
the value 3600 will be used.  The maximum response lifetime is 2147483647.
@end deffn

@deffn {VNC} {rt export @var{rt-list}} {}
@deffnx {VNC} {rt import @var{rt-list}} {}
@deffnx {VNC} {rt both @var{rt-list}} {}
Specify route target import and export lists.  @var{rt-list} is a
space-separated list of route targets, each element of which is
in one of the following forms:
@itemize
@item @var{IPv4-address}:@var{two-byte-integer}
@item @var{four-byte-autonomous-system-number}:@var{two-byte-integer}
@item @var{two-byte-autonomous-system-number}:@var{four-byte-integer}
@end itemize

The first form, @code{rt export}, specifies an @var{export rt-list}.
The @var{export rt-list} will be attached to routes originated by
NVEs in the NVE group when they are advertised via BGP. 
If the NVE group definition does not specify an @var{export rt-list},
then the default @var{export rt-list} is used.
If neither a group nor a default @var{export rt-list} is configured,
then no RT list will be sent; in turn, these routes will probably
not be processed
by receiving NVAs.

The second form, @code{rt import} specifies an @var{import rt-list},
which is a filter for incoming routes.
In order to be made available to NVEs in the group,
incoming BGP VPN and @w{ENCAP} @w{SAFI} (when @code{vnc
advertise-un-method encap-safi} is set) routes must have
RT lists that have at least one route target in common with the
group's @var{import rt-list}.

If the NVE group definition does not specify an import filter,
then the default @var{import rt-list} is used.
If neither a group nor a default @var{import rt-list} is configured,
there can be no RT intersections when receiving BGP routes and
therefore no incoming BGP routes will be processed for the group.

The third, @code{rt both}, is a shorthand way of specifying both
lists simultaneously, and is equivalent to @code{rt export @var{rt-list}}
followed by @code{rt import @var{rt-list}}.
@end deffn

@deffn {VNC} {export bgp|zebra route-map MAP-NAME}
Specify that the named route-map should be applied to routes
being exported to bgp or zebra. 
This paramter is used in conjunction with 
@ref{Configuring Export of Routes to Other Routing Protocols}.
This item is optional.
@end deffn

@deffn {VNC} {export bgp|zebra no route-map}
Specify that no route-map should be applied to routes
being exported to bgp or zebra. 
This paramter is used in conjunction with 
@ref{Configuring Export of Routes to Other Routing Protocols}.
This item is optional.
@end deffn

@deffn {VNC} {export bgp|zebra ipv4|ipv6 prefix-list LIST-NAME}
Specify that the named prefix-list filter should be applied to
routes being exported to bgp or zebra.
Prefix-lists for ipv4 and ipv6 are independent of each other. 
This paramter is used in conjunction with 
@ref{Configuring Export of Routes to Other Routing Protocols}.
This item is optional.
@end deffn

@deffn {VNC} {export bgp|zebra no ipv4|ipv6 prefix-list}
Specify that no prefix-list filter should be applied to
routes being exported to bgp or zebra. 
This paramter is used in conjunction with 
@ref{Configuring Export of Routes to Other Routing Protocols}.
This item is optional.
@end deffn

@c The following example shows two @code{vnc nve-group} definitions.  The first one,
@c ``group1'', applies to the IPV4 virtual-network route prefix 172.16/16.  It
@c sets the response lifetime to 200 seconds.  It defines a route target
@c import-export filter for the route targets 1000:1 and 1000:2
@c 
@c The second @code{vnc nve-group} definition, ``group2'', applies to the IPV6
@c underlay-network route prefix 10.0.2/24.  It defines the same response
@c lifetime and import-export filter as ``group1''.
@c 
@c @example
@c vnc nve-group group1
@c     prefix vn 172.16/16
@c     response-lifetime 200
@c     rt both 1000:1 1000:2
@c exit-vnc
@c 
@c vnc nve-group group2
@c     prefix un 10.0.2/24
@c     response-lifetime 200
@c     rt both 1000:1 1000:2
@c exit-vnc
@c @end example

@node VNC L2 Group Configuration
@subsection VNC L2 Group Configuration

The route targets advertised with prefixes and addresses registered by
an NVE are determined based on the NVE's associated VNC NVE Group
Configuration, @pxref{VNC NVE Group Configuration}.  Layer 2 (L2) Groups
are used to override the route targets for an NVE's Ethernet
registrations based on the Logical Network Identifier and label value.
A Logical Network Identifier is used to uniquely identify a logical
Ethernet segment and is conceptually similar to the Ethernet Segment
Identifier defined in @cite{RFC7432, BGP MPLS-Based Ethernet VPN}.  Both
the Logical Network Identifier and Label are passed to VNC via RFP
prefix and address registration.

Note that a corresponding NVE group configuration must be present, and
that other NVE associated configuration information, notably RD, is
not impacted by L2 Group Configuration.

@deffn {VNC} {vnc l2-group @var{name}} {}
Enter VNC configuration mode for defining the L2 group @var{name}.  
Use @code{exit} or @code{exit-vnc} to exit group configuration mode.

@example
vnc l2-group group1
  ... configuration commands
exit-vnc
@end example
@end deffn

@deffn {VNC} {no vnc l2-group @var{name}} {}
Delete the L2 group named @var{name}.
@end deffn

The following statements are valid in a L2 group definition:

@deffn {VNC} {logical-network-id @var{VALUE}}
Define the Logical Network Identifier with a value in the range of
0-4294967295 that identifies the logical Ethernet segment. 
@end deffn

@deffn {VNC} {labels @var{label-list}}
@deffnx {VNC} {no labels @var{label-list}}
Add or remove labels associated with the group.  @var{label-list} is a
space separated list of label values in the range of 0-1048575.
@end deffn

@deffn {VNC} {rt import @var{rt-target}} {}
@deffnx {VNC} {rt export @var{rt-target}} {}
@deffnx {VNC} {rt both @var{rt-target}} {}
Specify the route target import and export value associated with the
group. A complete definition of these parameters is given above,
@pxref{VNC NVE Group Configuration}.
@end deffn


@node Configuring Redistribution of Routes from Other Routing Protocols
@subsection Configuring Redistribution of Routes from Other Routing Protocols

Routes from other protocols (including BGP) can be provided to VNC (both
for RFP and for redistribution via BGP)
from three sources: the zebra kernel routing process;
directly from the main (default) unicast BGP RIB; or directly
from a designated BGP unicast exterior routing RIB instance.

The protocol named in the @code{vnc redistribute} command indicates
the route source:
@code{bgp-direct} routes come directly from the main (default)
unicast BGP RIB and are available for RFP and are redistributed via BGP;
@code{bgp-direct-to-nve-groups} routes come directly from a designated
BGP unicast routing RIB and are made available only to RFP;
and routes from other protocols come from the zebra kernel
routing process.
Note that the zebra process does not need to be active if
only @code{bgp-direct} or @code{bgp-direct-to-nve-groups} routes are used.

@subsubsection @code{zebra} routes

Routes originating from protocols other than BGP must be obtained
via the zebra routing process.
Redistribution of these routes into VNC does not support policy mechanisms
such as prefix-lists or route-maps.

@subsubsection @code{bgp-direct} routes

@code{bgp-direct} redistribution supports policy via
prefix lists and route-maps. This policy is applied to incoming
original unicast routes before the redistribution translations
(described below) are performed.

Redistribution of @code{bgp-direct} routes is performed in one of three
possible modes: @code{plain}, @code{nve-group}, or @code{resolve-nve}.
The default mode is @code{plain}.
These modes indicate the kind of translations applied to routes before
they are added to the VNC RIB.

In @code{plain} mode, the route's next hop is unchanged and the RD is set
based on the next hop.
For @code{bgp-direct} redistribution, the following translations are performed:
@itemize @bullet
@item
The VN address is set to the original unicast route's next hop address.
@item
The UN address is NOT set. (VN->UN mapping will occur via
ENCAP route or attribute, based on @code{vnc advertise-un-method}
setting, generated by the RFP registration of the actual NVE) 
@item
The RD is set to as if auto:vn:0 were specified (i.e.,
@var{rd_type=IP=1}:@var{IPv4-address=VN-address}:@var{two-byte-integer=0})
@item
The RT list is included in the extended community list copied from the
original unicast route (i.e., it must be set in the original unicast route).
@end itemize



In @code{nve-group} mode, routes are registered with VNC as
if they came from an NVE in the nve-group designated in the
@code{vnc redistribute nve-group} command. The following
translations are performed:

@itemize @bullet
@item
The next hop/VN address is set to the VN prefix configured for the
redistribute nve-group.
@item
The UN address is set to the UN prefix configured for the
redistribute nve-group.
@item
The RD is set to the RD configured for the redistribute nve-group.
@item
The RT list is set to the RT list configured for the redistribute nve-group.
If @code{bgp-direct} routes are being redistributed, 
any extended communities present in the original unicast route
will also be included.
@end itemize


In @code{resolve-nve} mode, the next hop of the original BGP route is
typically the address of an NVE connected router (CE) connected by one or
more NVEs.
Each of the connected NVEs will register, via RFP, a VNC host route
to the CE.
This mode may be though of as a mechanism to proxy RFP registrations
of BGP unicast routes on behalf of registering NVEs.

Multiple copies of the BGP route, one per matching NVE host route, will be
added to VNC.
In other words, for a given BGP unicast route, each instance of a
RFP-registered host route to the unicast route's next hop will result
in an instance of an imported VNC route.
Each such imported VNC route will have a prefix equal to the original
BGP unicast route's prefix, and a next hop equal to the next hop of the
matching RFP-registered host route.
If there is no RFP-registered host route to the next hop of the BGP unicast
route, no corresponding VNC route will be imported.

The following translations are applied:

@itemize @bullet
@item
The Next Hop is set to the next hop of the NVE route (i.e., the
VN address of the NVE).

@item
The extended community list in the new route is set to the 
union of:
@itemize @minus
@item
Any extended communities in the original BGP route
@item
Any extended communities in the NVE route
@item
An added route-origin extended community with the next hop of the
original BGP route
is added to the new route.
The value of the local administrator field defaults 5226 but may
be configured by the user via the @code{roo-ec-local-admin} parameter.
@end itemize

@item
The Tunnel Encapsulation attribute is set to the value of the Tunnel
Encapsulation attribute of the NVE route, if any.

@end itemize

@subsubsection @code{bgp-direct-to-nve-groups} routes

Unicast routes from the main or a designated instance of BGP
may be redistributed to VNC as bgp-direct-to-nve-groups routes. These
routes are NOT announced via BGP,
but they are made available for local RFP lookup in response to
queries from NVEs.

A non-main/default BGP instance is configured using the
@code{bgp multiple-instance} and @code{router bgp AS view NAME}
commands as described elsewhere in this document.

In order for a route in the unicast BGP RIB to be made
available to a querying NVE, there must already be, available to
that NVE, an (interior) VNC route matching the next hop address
of the unicast route.
When the unicast route is provided to the NVE, its next hop 
is replaced by the next hop of the corresponding
NVE. If there are multiple longest-prefix-match VNC routes,
the unicast route will be replicated for each.

There is currently no policy (prefix-list or route-map) support
for @code{bgp-direct-to-nve-groups} routes.

@subsubsection Redistribution Command Syntax

@deffn {VNC} {vnc redistribute ipv4|ipv6 bgp|bgp-direct|ipv6 bgp-direct-to-nve-groups|connected|kernel|ospf|rip|static} {}
@deffnx {VNC} {vnc redistribute ipv4|ipv6 bgp-direct-to-nve-groups view @var{VIEWNAME}} {}
@deffnx {VNC} {no vnc redistribute ipv4|ipv6 bgp|bgp-direct|bgp-direct-to-nve-groups|connected|kernel|ospf|rip|static} {}
Import (or do not import) prefixes from another routing
protocols. Specify both the address family to import (@code{ipv4} or
@code{ipv6}) and the protocol (@code{bgp}, @code{bgp-direct},
@code{bgp-direct-to-nve-groups}, @code{connected},
@code{kernel}, @code{ospf}, @code{rip}, or @code{static}).  Repeat
this statement as needed for each combination of address family and
routing protocol.
Prefixes from protocol @code{bgp-direct} are imported from unicast BGP
in the same bgpd process.
Prefixes from all other protocols (including @code{bgp}) are imported
via the @code{zebra} kernel routing process.
@end deffn

@deffn {VNC} {vnc redistribute mode plain|nve-group|resolve-nve}
Redistribute routes from other protocols into VNC using the
specified mode.
Not all combinations of modes and protocols are supported.
@end deffn

@deffn {VNC} {vnc redistribute nve-group @var{group-name}} {}
@deffnx {VNC} {no vnc redistribute nve-group @var{group-name}} {}
When using @code{nve-group} mode,
assign (or do not assign) the NVE group @var{group-name} to routes
redistributed from another routing protocol.  @var{group-name}
must be configured using @code{vnc nve-group}.

The VN and UN prefixes of the nve-group must both be configured,
and each prefix must be specified as a full-length (/32 for IPv4,
/128 for IPv6) prefix.
@end deffn

@deffn {VNC} {vnc redistribute lifetime @var{lifetime}|infinite} {}
Assign a registration lifetime, either @var{lifetime} seconds or
@code{infinite}, to prefixes redistributed from other routing
protocols as if they had been received via RFP registration messages
from an NVE.  @var{lifetime} can be any integer between 1 and
4294967295, inclusive. 
@end deffn

@deffn {VNC} {vnc redistribute resolve-nve roo-ec-local-admin @var{0-65536}}
Assign a value to the local-administrator subfield used in the
Route Origin extended community that is assigned to routes exported 
under the @code{resolve-nve} mode. The default value is @var{5226}.
@end deffn

The following four @code{prefix-list} and @code{route-map} commands
may be specified in the context of an nve-group or not.
If they are specified in the context of an nve-group, they
apply only if the redistribution mode is @code{nve-group},
and then only for routes being redistributed from
@code{bgp-direct}.
If they are specified outside the context of an nve-group, then
they apply only for redistribution modes @code{plain} and @code{resolve-nve},
and then only for routes being redistributed from @code{bgp-direct}.

@deffn {VNC} {vnc redistribute bgp-direct (ipv4|ipv6) prefix-list @var{LIST-NAME}}
When redistributing @code{bgp-direct} routes,
specifies that the named prefix-list should be applied.
@end deffn

@deffn {VNC} {vnc redistribute bgp-direct no (ipv4|ipv6) prefix-list}
When redistributing @code{bgp-direct} routes,
specifies that no prefix-list should be applied.
@end deffn

@deffn {VNC} {vnc redistribute bgp-direct route-map  @var{MAP-NAME}}
When redistributing @code{bgp-direct} routes,
specifies that the named route-map should be applied.
@end deffn

@deffn {VNC} {vnc redistribute bgp-direct no route-map}
When redistributing @code{bgp-direct} routes,
specifies that no route-map should be applied.
@end deffn

@node Configuring Export of Routes to Other Routing Protocols
@subsection Configuring Export of Routes to Other Routing Protocols

Routes from VNC (both for RFP and for redistribution via BGP) can be
provided to other protocols, either via zebra or directly to BGP.

It is important to note that when exporting routes to other protocols,
the downstream protocol must also be configured to import the routes.
For example, when VNC routes are exported to unicast BGP, the BGP
configuration must include a corresponding @code{redistribute vnc-direct}
statement.

@deffn {VNC} {export bgp|zebra mode none|group-nve|registering-nve|ce}
Specify how routes should be exported to bgp or zebra.
If the mode is @code{none}, routes are not exported.
If the mode is @code{group-nve}, routes are exported according
to nve-group configuration (@pxref{VNC NVE Group Configuration}): if a group is configured to
allow export, then each prefix visible to the group is exported
with next hops set to the currently-registered NVEs.
If the mode is @code{registering-nve}, then all VNC routes are
exported with their original next hops.
If the mode is @code{ce}, only VNC routes that have an NVE connected CE Router
encoded in a Route Origin Extended Community are exported.
This extended community must have an administrative value that
matches the configured @code{roo-ec-local-admin} value.
The next hop of the exported route is set to the encoded
NVE connected CE Router.

The default for both bgp and zebra is mode @code{none}.
@end deffn

@deffn {VNC} {vnc export bgp|zebra group-nve group @var{group-name}}
@deffnx {VNC} {vnc export bgp|zebra group-nve no group @var{group-name}}
When export mode is @code{group-nve},
export (or do not export) prefixes from the specified nve-group
to unicast BGP or to zebra.
Repeat this statement as needed for each nve-group to be exported.
Each VNC prefix that is exported will result in N exported routes to the
prefix, each with a next hop corresponding to one of the N NVEs currently
associated with the nve-group.
@end deffn

@deffn {VNC} export bgp|zebra ipv4|ipv6 prefix-list LIST-NAME
When export mode is @code{ce} or @code{registering-nve},
specifies that the named prefix-list should be applied to routes
being exported to bgp or zebra.
Prefix-lists for ipv4 and ipv6 are independent of each other.
@end deffn

@deffn {VNC} export bgp|zebra no ipv4|ipv6 prefix-list
When export mode is @code{ce} or @code{registering-nve},
specifies that no prefix-list should be applied to routes
being exported to bgp or zebra.
@end deffn

@deffn {VNC} export bgp|zebra route-map MAP-NAME
When export mode is @code{ce} or @code{registering-nve},
specifies that the named route-map should be applied to routes
being exported to bgp or zebra.
@end deffn

@deffn {VNC} export bgp|zebra no route-map
When export mode is @code{ce} or @code{registering-nve},
specifies that no route-map should be applied to routes
being exported to bgp or zebra.
@end deffn

When the export mode is @code{group-nve}, policy for exported
routes is specified per-NVE-group inside a @code{nve-group} @var{RFG-NAME} block
via the following commands(@pxref{VNC NVE Group Configuration}):

@deffn {VNC} {export bgp|zebra route-map MAP-NAME}
This command is valid inside a @code{nve-group} @var{RFG-NAME} block.
It specifies that the named route-map should be applied to routes
being exported to bgp or zebra.
@end deffn

@deffn {VNC} {export bgp|zebra no route-map}
This command is valid inside a @code{nve-group} @var{RFG-NAME} block.
It specifies that no route-map should be applied to routes
being exported to bgp or zebra.
@end deffn

@deffn {VNC} {export bgp|zebra ipv4|ipv6 prefix-list LIST-NAME}
This command is valid inside a @code{nve-group} @var{RFG-NAME} block.
It specifies that the named prefix-list filter should be applied to
routes being exported to bgp or zebra.
Prefix-lists for ipv4 and ipv6 are independent of each other.
@end deffn

@deffn {VNC} {export bgp|zebra no ipv4|ipv6 prefix-list}
This command is valid inside a @code{nve-group} @var{RFG-NAME} block.
It specifies that no prefix-list filter should be applied to
routes being exported to bgp or zebra.
@end deffn

@node Manual Address Control
@section Manual Address Control

The commands in this section can be used to augment normal dynamic VNC.
The @code{add vnc} commands can be used to manually add IP prefix or
Ethernet MAC address forwarding information.  The @code{clear vnc}
commands can be used to remove manually and dynamically added
information.

@deffn {Command} {add vnc prefix (A.B.C.D/M|X:X::X:X/M) vn (A.B.C.D|X:X::X:X) un (A.B.C.D|X:X::X:X) [cost <0-255>] [lifetime (infinite|<1-4294967295>)] [local-next-hop (A.B.C.D|X:X::X:X) [local-cost <0-255>]]} {}
Register an IP prefix on behalf of the NVE identified by the VN and UN
addresses.  The @code{cost} parameter provides the administrative
preference of the forwarding information for remote advertisement.  If
omitted, it defaults to 255 (lowest preference).  The @code{lifetime}
parameter identifies the period, in seconds, that the information
remains valid.  If omitted, it defaults to @var{infinite}.  The optional
@code{local-next-hop} parameter is used to configure a nexthop to be
used by an NVE to reach the prefix via a locally connected CE router.
This information remains local to the NVA, i.e., not passed to other
NVAs, and is only passed to registered NVEs. When specified, it is also
possible to provide a @code{local-cost} parameter to provide a
forwarding preference.  If omitted, it defaults to 255 (lowest
preference).
@end deffn


@deffn {Command} {add vnc mac xx:xx:xx:xx:xx:xx virtual-network-identifier <1-4294967295> vn (A.B.C.D|X:X::X:X) un (A.B.C.D|X:X::X:X) [prefix (A.B.C.D/M|X:X::X:X/M)] [cost <0-255>] [lifetime (infinite|<1-4294967295>)]} {}
Register a MAC address for a logical Ethernet (L2VPN) on behalf of the
NVE identified by the VN and UN addresses.
The optional @code{prefix} parameter is to support enable IP address
mediation for the given prefix.   The @code{cost} parameter provides the administrative
preference of the forwarding information.  If omitted, it defaults to
255.  The @code{lifetime} parameter identifies the period, in seconds,
that the information remains valid.  If omitted, it defaults to
@var{infinite}. 
@end deffn

@deffn {Command} {clear vnc prefix (*|A.B.C.D/M|X:X::X:X/M) (*|[(vn|un) (A.B.C.D|X:X::X:X|*) [(un|vn) (A.B.C.D|X:X::X:X|*)] [mac xx:xx:xx:xx:xx:xx] [local-next-hop (A.B.C.D|X:X::X:X)])} {}
Delete the information identified by prefix, VN address, and UN address.
Any or all of these parameters may be wilcarded to (potentially) match
more than one registration.
The optional @code{mac} parameter specifies a layer-2 MAC address
that must match the registration(s) to be deleted.
The optional @code{local-next-hop} parameter is used to
delete specific local nexthop information.
@end deffn

@deffn {Command} {clear vnc mac (*|xx:xx:xx:xx:xx:xx) virtual-network-identifier (*|<1-4294967295>) (*|[(vn|un) (A.B.C.D|X:X::X:X|*) [(un|vn) (A.B.C.D|X:X::X:X|*)] [prefix (*|A.B.C.D/M|X:X::X:X/M)])} {}
Delete mac forwarding information.
Any or all of these parameters may be wilcarded to (potentially) match
more than one registration.
The default value for the @code{prefix} parameter is the wildcard value @var{*}.
@end deffn

@deffn {Command} {clear vnc nve (*|((vn|un) (A.B.C.D|X:X::X:X) [(un|vn) (A.B.C.D|X:X::X:X)])) } {}
Delete prefixes associated with the NVE specified by the given VN and UN
addresses.
It is permissible to specify only one of VN or UN, in which case
any matching registration will be deleted.
It is also permissible to specify @code{*} in lieu of any VN or UN
address, in which case all registrations will match.
@end deffn

@node Other VNC-Related Commands
@section Other VNC-Related Commands

Note: VNC-Related configuration can be obtained via the @code{show
running-configuration} command when in @code{enable} mode.

The following commands are used to clear and display 
Virtual Network Control related information:

@deffn {COMMAND} {clear vnc counters} {}
Reset the counter values stored by the NVA. Counter
values can be seen using the @code{show vnc} commands listed above. This
command is only available in @code{enable} mode.
@end deffn

@deffn {Command} {show vnc summary} {}
Print counter values and other general information 
about the NVA. Counter values can be reset 
using the @code{clear vnc counters} command listed below.
@end deffn

@deffn {Command} {show vnc nves} {}
@deffnx {Command} {show vnc nves vn|un @var{address}} {}
Display the NVA's current clients. Specifying @var{address}
limits the output to the NVEs whose addresses match @var{address}.
The time since the NVA last communicated with the NVE, per-NVE
summary counters and each NVE's addresses will be displayed.
@end deffn

@deffn {Command} {show vnc queries} {}
@deffnx {Command} {show vnc queries @var{prefix}} {}
Display active Query information.  Queries remain valid for the default
Response Lifetime (@pxref{VNC Defaults Configuration}) or NVE-group
Response Lifetime (@pxref{VNC NVE Group Configuration}).  Specifying
@var{prefix} limits the output to Query Targets that fall within
@var{prefix}.

Query information is provided for each querying NVE, and includes the
Query Target and the time remaining before the information is removed.
@end deffn

@deffn {Command} {show vnc registrations [all|local|remote|holddown|imported]} {}
@deffnx {Command} {show vnc registrations [all|local|remote|holddown|imported] @var{prefix}} {}
Display local, remote, holddown, and/or imported registration information.
Local registrations are routes received via RFP, which are present in the
NVA Registrations Cache.
Remote registrations are routes received via BGP (VPN SAFIs), which
are present in the NVE-group import tables.
Holddown registrations are local and remote routes that have been
withdrawn but whose holddown timeouts have not yet elapsed.
Imported information represents routes that are imported into NVA and
are made available to querying NVEs.  Depending on configuration,
imported routes may also be advertised via BGP.
Specifying @var{prefix} limits the output to the registered prefixes that
fall within @var{prefix}.

Registration information includes the registered prefix, the registering
NVE addresses, the registered administrative cost, the registration
lifetime and the time since the information was registered or, in the
case of Holddown registrations, the amount of time remaining before the
information is removed.
@end deffn

@deffn {Command} {show vnc responses [active|removed]} {}
@deffnx {Command} {show vnc responses [active|removed] @var{prefix}} {}
Display all, active and/or removed response information which are
present in the NVA Responses Cache. Responses remain valid for the
default Response Lifetime (@pxref{VNC Defaults Configuration}) or
NVE-group Response Lifetime (@pxref{VNC NVE Group Configuration}.)
When Removal Responses are enabled (@pxref{General VNC Configuration}),
such responses are listed for the Response Lifetime.  Specifying
@var{prefix} limits the output to the addresses that fall within
@var{prefix}.

Response information is provided for each querying NVE, and includes
the response prefix, the prefix-associated registering NVE addresses,
the administrative cost, the provided response lifetime and the time
remaining before the information is to be removed or will become inactive.
@end deffn

@deffn {Command} {show memory vnc} {}
Print the number of memory items allocated by the NVA.
@end deffn

@node Example VNC and VNC-GW Configurations
@section Example VNC and VNC-GW Configurations

@menu
* Mesh NVA Configuration::
* Mesh NVA and VNC-GW Configuration::
* VNC with Frr Route Reflector Configuration::
* VNC with Commercial Route Reflector Configuration::
* VNC with Redundant Route Reflectors Configuration::
@c * Interfacing VNC to an IGP::
@end menu

@node  Mesh NVA Configuration
@subsection  Mesh NVA Configuration

This example includes three NVAs, nine NVEs, and two NVE groups. Note
that while not shown, a single physical device may support multiple
logical NVEs.  @ref{fig:fig-vnc-mesh} shows @code{NVA 1}
(192.168.1.100), @code{NVA 2} (192.168.1.101), and @code{NVA 3}
(192.168.1.102), which are connected in a full mesh.  Each is a
member of the autonomous system 64512.  Each NVA provides VNC
services to three NVE clients in the 172.16.0.0/16 virtual-network
address range.  The 172.16.0.0/16 address range is partitioned into
two NVE groups, @code{group1} (172.16.0.0/17) and @code{group2}
(172.16.128.0/17).

Each NVE belongs to either NVE group @code{group1} or NVE group
@code{group2}.  The NVEs @code{NVE 1}, @code{NVE 2}, @code{NVE
4}, @code{NVE 7}, and @code{NVE 8} are members of the NVE group
@code{group1}.  The NVEs @code{NVE 3}, @code{NVE 5}, @code{NVE
6}, and @code{NVE 9} are members of the NVE group @code{group2}.

Each NVA advertises NVE underlay-network IP addresses using the
Tunnel Encapsulation Attribute.

@float Figure,fig:fig-vnc-mesh
@center @image{fig-vnc-mesh,400pt,,Three-way Mesh}
@caption{A three-way full mesh with three NVEs per NVA}
@end float

@file{bgpd.conf} for @code{NVA 1} (192.168.1.100)
@verbatim
router bgp 64512

    bgp router-id 192.168.1.100

    neighbor 192.168.1.101  remote-as 64512
    neighbor 192.168.1.102  remote-as 64512

    address-family vpnv4
        neighbor 192.168.1.101 activate
        neighbor 192.168.1.102 activate
    exit-address-family

    vnc defaults
        rd 64512:1
        response-lifetime 200
        rt both 1000:1 1000:2
    exit-vnc

    vnc nve-group group1
        prefix vn 172.16.0.0/17
        rt both 1000:1 
    exit-vnc

    vnc nve-group group2
        prefix vn 172.16.128.0/17
        rt both 1000:2
    exit-vnc

exit
@end verbatim

@file{bgpd.conf} for @code{NVA 2} (192.168.1.101):
@verbatim
router bgp 64512

    bgp router-id 192.168.1.101

    neighbor 192.168.1.100  remote-as 64512
    neighbor 192.168.1.102  remote-as 64512

    address-family vpnv4
        neighbor 192.168.1.100 activate
        neighbor 192.168.1.102 activate
    exit-address-family

    vnc nve-group group1
        prefix vn 172.16.0.0/17
        rd 64512:1
        response-lifetime 200
        rt both 1000:1 1000:2
    exit-vnc
exit
@end verbatim

@file{bgpd.conf} for @code{NVA 3} (192.168.1.102):
@verbatim
router bgp 64512

    bgp router-id 192.168.1.102

    neighbor 192.168.1.101  remote-as 64512
    neighbor 192.168.1.102  remote-as 64512

    address-family vpnv4
        neighbor 192.168.1.100 activate
        neighbor 192.168.1.101 activate
    exit-address-family

    vnc defaults
        rd 64512:1
        response-lifetime 200
        rt both 1000:1 1000:2
    exit-vnc

    vnc nve-group group1
        prefix vn 172.16.128.0/17
    exit-vnc
exit
@end verbatim

@node Mesh NVA and VNC-GW Configuration
@subsection Mesh NVA and VNC-GW Configuration

This example includes two NVAs, each with two associated NVEs, and two
VNC-GWs, each supporting two CE routers physically attached to the four
NVEs.  Note that this example is showing a more complex configuration
where VNC-GW is separated from normal NVA functions; it is equally
possible to simplify the configuration and combine NVA and VNC-GW
functions in a single frr instance.

@float Figure,fig:fig-vnc-gw
@center @image{fig-vnc-gw,400pt,,Frr VNC Gateway}
@caption{Meshed NVEs and VNC-GWs}
@end float

As shown in @ref{fig:fig-vnc-gw}, NVAs and VNC-GWs are connected in a
full iBGP mesh.  The VNC-GWs each have two CEs configured as
route-reflector clients.  Each client provides BGP updates with unicast
routes that the VNC-GW reflects to the other client.  The VNC-GW also
imports these unicast routes into VPN routes to be shared with the other
VNC-GW and the two NVAs.  This route importation is controlled with the
@code{vnc redistribute} statements shown in the configuration.
Similarly, registrations sent by NVEs via RFP to the NVAs are exported
by the VNC-GWs to the route-reflector clients as unicast routes.  RFP
registrations exported this way have a next-hop address of the CE behind
the connected (registering) NVE.  Exporting VNC routes as IPv4 unicast
is enabled with the @code{vnc export} command below.

The configuration for @code{VNC-GW 1} is shown below.
@verbatim
router bgp 64512
 bgp router-id 192.168.1.101
 bgp cluster-id 1.2.3.4
 neighbor 192.168.1.102 remote-as 64512
 neighbor 192.168.1.103 remote-as 64512
 neighbor 192.168.1.104 remote-as 64512
 neighbor 172.16.1.2 remote-as 64512
 neighbor 172.16.2.2 remote-as 64512
 !
 address-family ipv4 unicast
  redistribute vnc-direct
  no neighbor 192.168.1.102 activate
  no neighbor 192.168.1.103 activate
  no neighbor 192.168.1.104 activate
  neighbor 172.16.1.2 route-reflector-client
  neighbor 172.16.2.2 route-reflector-client
 exit-address-family
 !
 address-family vpnv4 unicast
   neighbor 192.168.1.102 activate
   neighbor 192.168.1.103 activate
   neighbor 192.168.1.104 activate
 exit-address-family
 vnc export bgp mode ce
 vnc redistribute mode resolve-nve
 vnc redistribute ipv4 bgp-direct
 exit
@end verbatim

Note that in the VNC-GW configuration, the neighboring VNC-GW and
NVAs each have a statement disabling the IPv4 unicast address family.
IPv4 unicast is on by default and this prevents the other VNC-GW and
NVAs from learning unicast routes advertised by the route-reflector clients.

Configuration for @code{NVA 2}:
@verbatim
router bgp 64512
 bgp router-id 192.168.1.104
 neighbor 192.168.1.101 remote-as 64512
 neighbor 192.168.1.102 remote-as 64512
 neighbor 192.168.1.103 remote-as 64512
 !
 address-family ipv4 unicast
  no neighbor 192.168.1.101 activate
  no neighbor 192.168.1.102 activate
  no neighbor 192.168.1.103 activate
 exit-address-family
 !
 address-family vpnv4 unicast
   neighbor 192.168.1.101 activate
   neighbor 192.168.1.102 activate
   neighbor 192.168.1.103 activate
 exit-address-family
 !
 vnc defaults
  response-lifetime 3600
  exit-vnc
 vnc nve-group nve1
  prefix vn 172.16.1.1/32
  response-lifetime 3600
  rt both 1000:1 1000:2
  exit-vnc
 vnc nve-group nve2
  prefix vn 172.16.2.1/32
  response-lifetime 3600
  rt both 1000:1 1000:2
  exit-vnc
 exit
@end verbatim

@c TBD make this its own example:
@c 
@c @float Figure,fig:fig-vnc-gw-rr
@c @center @image{fig-vnc-gw-rr,400pt,,Frr VNC Gateway with RR}
@c @end float
@c An NVA can also import unicast routes from BGP without advertising the
@c imported routes as VPN routes.  Such imported routes, while not
@c distributed to other NVAs or VNC-GWs, are are available to NVEs via
@c RFP query messages sent to the NVA. @ref{fig:fig-vnc-gw-rr}
@c shows an example topology where unicast routes are imported into NVAs
@c from a Route Reflector.  (@pxref{Route Reflector} for route reflector
@c configuration details.)  The following three lines can be added to the
@c @code{NVA 1} and @code{NVA 2} configurations to import routes into VNC
@c for local VNC use:
@c 
@c @verbatim
@c  neighbor 192.168.1.105 remote-as 64512
@c  vnc redistribute mode plain
@c  vnc redistribute ipv4 bgp-direct-to-nve-groups
@c @end verbatim

@node VNC with Frr Route Reflector Configuration
@subsection VNC with Frr Route Reflector Configuration
A route reflector eliminates the need for a fully meshed NVA
network by acting as the hub between NVAs.
@ref{fig:fig-vnc-frr-route-reflector} shows BGP route reflector
@code{BGP Route Reflector 1} (192.168.1.100) as a route reflector for
NVAs @code{NVA 2}(192.168.1.101) and @code{NVA 3}
(192.168.1.102).

@float Figure,fig:fig-vnc-frr-route-reflector
@center @image{fig-vnc-frr-route-reflector,400pt,,Frr Route Reflector}
@caption{Two NVAs and a BGP Route Reflector} 
@end float

@code{NVA 2} and @code{NVA 3} 
advertise NVE underlay-network IP addresses using the Tunnel Encapsulation Attribute.
@code{BGP Route Reflector 1} ``reflects'' advertisements from
@code{NVA 2} to @code{NVA 3} and vice versa.

As in the example of @ref{Mesh NVA Configuration}, there are two NVE groups.
The 172.16.0.0/16 address range is partitioned into two NVE groups,
@code{group1} (172.16.0.0/17) and @code{group2} (172.16.128.0/17).
The NVE @code{NVE 4}, @code{NVE 7}, and @code{NVE 8} are
members of the NVE group @code{group1}.  The NVEs @code{NVE 5},
@code{NVE 6}, and @code{NVE 9} are members of the NVE group
@code{group2}.

@file{bgpd.conf} for @code{BGP Route Reflector 1} on 192.168.1.100:
@verbatim
router bgp 64512

    bgp router-id 192.168.1.100

    neighbor 192.168.1.101 remote-as 64512
    neighbor 192.168.1.101 port 7179
    neighbor 192.168.1.101 description iBGP-client-192-168-1-101

    neighbor 192.168.1.102 remote-as 64512
    neighbor 192.168.1.102 port 7179
    neighbor 192.168.1.102 description iBGP-client-192-168-1-102

    address-family ipv4 unicast
        neighbor 192.168.1.101 route-reflector-client
        neighbor 192.168.1.102 route-reflector-client
    exit-address-family

    address-family vpnv4
        neighbor 192.168.1.101 activate
        neighbor 192.168.1.102 activate

        neighbor 192.168.1.101 route-reflector-client
        neighbor 192.168.1.102 route-reflector-client
    exit-address-family

exit
@end verbatim

@file{bgpd.conf} for @code{NVA 2} on 192.168.1.101:
@verbatim
router bgp 64512

    bgp router-id 192.168.1.101

    neighbor 192.168.1.100  remote-as 64512

    address-family vpnv4
        neighbor 192.168.1.100 activate
    exit-address-family

    vnc nve-group group1
        prefix vn 172.16.0.0/17
        rd 64512:1
        response-lifetime 200
        rt both 1000:1 1000:2
    exit-vnc
exit
@end verbatim

@file{bgpd.conf} for @code{NVA 2} on 192.168.1.102:
@verbatim
router bgp 64512

    bgp router-id 192.168.1.102

    neighbor 192.168.1.100  remote-as 64512

    address-family vpnv4
        neighbor 192.168.1.100 activate
    exit-address-family

    vnc defaults
        rd 64512:1
        response-lifetime 200
        rt both 1000:1 1000:2
    exit-vnc

    vnc nve-group group1
        prefix vn 172.16.128.0/17
    exit-vnc
exit
@end verbatim

While not shown, an NVA can also be configured as a route reflector.

@node VNC with Commercial Route Reflector Configuration
@subsection VNC with Commercial Route Reflector Configuration
This example is identical to @ref{VNC with Frr Route Reflector
Configuration} with the exception that the route reflector is a
commercial router.  Only the
VNC-relevant configuration is provided. 

@float Figure,fig:fig-vnc-commercial-route-reflector
@center @image{fig-vnc-commercial-route-reflector,400pt,,Commercial Route Reflector}
@caption{Two NVAs with a commercial route reflector}
@end float

@file{bgpd.conf} for BGP route reflector @code{Commercial Router} on 192.168.1.104:
@verbatim
version 8.5R1.13;
routing-options {
    rib inet.0 {
        static {
            route 172.16.0.0/16 next-hop 192.168.1.104;
        }
    }
    autonomous-system 64512;
    resolution {
        rib inet.3 {
            resolution-ribs inet.0;
        }
        rib bgp.l3vpn.0 {
            resolution-ribs inet.0;
        }
    }
}
protocols {
    bgp {
        advertise-inactive;
        family inet {
            labeled-unicast;
        }
       group 1 {
            type internal;
            advertise-inactive;
            advertise-peer-as;
            import h;
            family inet {
                unicast;
            }
            family inet-vpn {
                unicast;
            }
            cluster 192.168.1.104;
            neighbor 192.168.1.101;
            neighbor 192.168.1.102;
        }
    }
}
policy-options {
    policy-statement h {
        from protocol bgp;
        then {
            as-path-prepend 64512;
            accept;
        }
    }
}
@end verbatim

@file{bgpd.conf} for @code{NVA 2} on 192.168.1.101:
@verbatim
router bgp 64512

    bgp router-id 192.168.1.101

    neighbor 192.168.1.100  remote-as 64512

    address-family vpnv4
        neighbor 192.168.1.100 activate
    exit-address-family

    vnc nve-group group1
        prefix vn 172.16.0.0/17
        rd 64512:1
        response-lifetime 200
        rt both 1000:1 1000:2
    exit-vnc
exit
@end verbatim

@file{bgpd.conf} for @code{NVA 3} on 192.168.1.102:
@verbatim
router bgp 64512

    bgp router-id 192.168.1.102

    neighbor 192.168.1.100  remote-as 64512

    address-family vpnv4
        neighbor 192.168.1.100 activate
    exit-address-family

    vnc defaults
        rd 64512:1
        response-lifetime 200
        rt both 1000:1 1000:2
    exit-vnc

    vnc nve-group group1
        prefix vn 172.16.128.0/17
    exit-vnc
exit
@end verbatim

@node VNC with Redundant Route Reflectors Configuration
@subsection VNC with Redundant Route Reflectors Configuration
This example combines the previous two (@ref{VNC with Frr Route
Reflector Configuration} and @ref{VNC with Commercial Route Reflector
Configuration}) into a redundant route reflector configuration.  BGP
route reflectors @code{BGP Route Reflector 1} and @code{Commercial Router}
are the route reflectors for NVAs @code{NVA 2} and
@code{NVA 3}.  The two NVAs have connections to both
route reflectors.

@float Figure,fig:fig-vnc-redundant-route-reflectors
@center @image{fig-vnc-redundant-route-reflectors,400pt,,Redundant Route Reflectors}
@caption{Frr-based NVA with redundant route reflectors}
@end float

@file{bgpd.conf} for @code{Bgpd Route Reflector 1} on 192.168.1.100:
@verbatim
router bgp 64512

    bgp router-id 192.168.1.100
    bgp cluster-id 192.168.1.100

    neighbor 192.168.1.104 remote-as 64512

    neighbor 192.168.1.101 remote-as 64512
    neighbor 192.168.1.101 description iBGP-client-192-168-1-101
    neighbor 192.168.1.101 route-reflector-client

    neighbor 192.168.1.102 remote-as 64512
    neighbor 192.168.1.102 description iBGP-client-192-168-1-102
    neighbor 192.168.1.102 route-reflector-client

    address-family vpnv4
        neighbor 192.168.1.101 activate
        neighbor 192.168.1.102 activate
        neighbor 192.168.1.104 activate

        neighbor 192.168.1.101 route-reflector-client
        neighbor 192.168.1.102 route-reflector-client
    exit-address-family
exit
@end verbatim

@file{bgpd.conf} for @code{NVA 2} on 192.168.1.101:
@verbatim
router bgp 64512

    bgp router-id 192.168.1.101

    neighbor 192.168.1.100  remote-as 64512
    neighbor 192.168.1.104  remote-as 64512

    address-family vpnv4
        neighbor 192.168.1.100 activate
        neighbor 192.168.1.104 activate
    exit-address-family

    vnc nve-group group1
        prefix vn 172.16.0.0/17
        rd 64512:1
        response-lifetime 200
        rt both 1000:1 1000:2
    exit-vnc
exit
@end verbatim

@file{bgpd.conf} for @code{NVA 3} on 192.168.1.102:
@verbatim
router bgp 64512

    bgp router-id 192.168.1.102

    neighbor 192.168.1.100  remote-as 64512
    neighbor 192.168.1.104  remote-as 64512

    address-family vpnv4
        neighbor 192.168.1.100 activate
        neighbor 192.168.1.104 activate
    exit-address-family

    vnc defaults
        rd 64512:1
        response-lifetime 200
        rt both 1000:1 1000:2
    exit-vnc

    vnc nve-group group1
        prefix vn 172.16.128.0/17
    exit-vnc
exit
@end verbatim

@file{bgpd.conf} for the Commercial Router route reflector on
192.168.1.104:
@verbatim
routing-options {
    rib inet.0 {
        static {
            route 172.16.0.0/16 next-hop 192.168.1.104;
        }
    }
    autonomous-system 64512;
    resolution {
        rib inet.3 {
            resolution-ribs inet.0;
        }
        rib bgp.l3vpn.0 {
            resolution-ribs inet.0;
        }
    }
}
protocols {
    bgp {
        advertise-inactive;
        family inet {
            labeled-unicast;
        }
       group 1 {
            type internal;
            advertise-inactive;
            advertise-peer-as;
            import h;
            family inet {
                unicast;
            }
            family inet-vpn {
                unicast;
            }
            cluster 192.168.1.104;
            neighbor 192.168.1.101;
            neighbor 192.168.1.102;
        }

       group 2 {
            type internal;
            advertise-inactive;
            advertise-peer-as;
            import h;
            family inet {
                unicast;
            }
            family inet-vpn {
                unicast;
            }
            neighbor 192.168.1.100;
        }

    }
}
policy-options {
    policy-statement h {
        from protocol bgp;
        then {
            as-path-prepend 64512;
            accept;
        }
    }
}
@end verbatim

@node Release Notes
@section Release Notes

@c A paragraph that introduces our release notes.

@c outer list, one item per VNC release, items preceded by bullet
@itemize @bullet 
@item

@c @item
@end itemize

@evenheading @thispage@|@|@thistitle
@oddheading @thischapter@|@|@thispage
@everyfooting