User's Manual

Configure VRF-lite | Page 47
Configuring a complex inter-VRF solution
Local interfaces can be utilised by a number of protocols for various purposes. They can be
used as a reliable address via which to access a device - an address that is always accessible,
irrespective of the link status of any individual external interface.
Within each VRF, configure optional route distinguisher (RD), route-targets and VRF import
and export maps. The RD, route-targets and VRF import and export maps are used when
leaking routes via BGP. They are not required when inter-VRF communication is achieved via
static inter-VRF routes. BGP is used to facilitate inter-VRF communication in this example.
The RD is a BGP ASN (xxx:). The VRF RD is also used by MPLS to facilitate VRF VPNs,
which are currently not supported, and thus serves little purpose in the context of VRF-lite.
However the RD command is required if using BGP to facilitate inter-VRF communications
.
Each RD references a unique VRF instance (:xxx). A complete VRF ASN uses the syntax
xxx:xxx. For example 100:1
denotes BGP ASN 100, VRF instance1.
The command route-target export xxx:xxx enables routes in the VRF domain with a
matching VRF ASN tag to be exported via BGP to be subsequently leaked to other VRFs.
The command route-target import xxx:xxx enables routes from other VRF domains with a
matching VRF ASN tag to be imported via BGP into the VRF domain.
The command export map name references a route map, which in turn references the ACLs
previously configured. This command ensures (via the associated ACLs) that only selected
routes are exported from the VRF domain to BGP.
In this example VRF orange has a static route to network 192.168.20.0/24. This same IP
subnet is assigned to vlan 2, which is a part of VRF green. Therefore there is an export map
(orange 140) and associated ACL orangeBlock20Export140 deny 192.168.20.0/24 to ensure
the network 192.168.20.0 is not exported into BGP, whilst still allowing the export of other
networks that do not match the ACL.
The command import map name references a route map, which in turn references the ACLs
previously configured. This commands ensures (via the associated ACLs) that only selected
routes are imported into the VRF domain from BGP.
There is no route leakage to or from VRF overlap. VRF overlap (and its associated VLANs)
remain completely isolated from all other VRF domains. VRF overlap contains network
192.168.10.0/24 associated with vlan6. This same subnet is also contained in VRF red vlan1.
This is OK, as VRF overlap has no associated route-target import and route-target export
commands.