This example is designed to show how a sham-link is used only to affect the OSPF intra-area path selection of the PE and CE routers. The PE router also uses the information received from MP-BGP to set the outgoing label stack of incoming packets, and to decide to which egress PE router to label switch the packets.
The figure below shows a sample MPLS VPN topology in which a sham-link configuration is necessary. A VPN client has three sites, each with a backdoor link. Two sham-links have been configured, one between PE-1 and PE-2, and another between PE-2 and PE-3. A sham-link between PE-1 and PE-3 is not necessary in this configuration because the Vienna and Winchester sites do not share a backdoor link.
The following output shows the forwarding that occurs between sites from the standpoint of how PE-1 views the 10.3.1.7/32 prefix, the loopback1 interface of the Winchester CE router in the figure.
PE-1# show ip bgp vpnv4 all 10.3.1.7
BGP routing table entry for 100:251:10.3.1.7/32, version 124
Paths: (1 available, best #1)
Local
10.3.1.2 (metric 30) from 10.3.1.2
(10.3.1.2)
Origin incomplete, metric 11, localpref 100, valid, internal,
best
Extended Community: RT:1:793 OSPF DOMAIN ID:0.0.0.100 OSPF
RT:1:2:0 OSPF 2
PE-1# show ip route vrf ospf 10.3.1.7
Routing entry for 10.3.1.7/32
Known via "ospf 100
", distance 110, metric 13, type intra area
Redistributing via bgp 215
Last update from 10.3.1.2 00:12:59 ago
Routing Descriptor Blocks:
10.3.1.2 (Default-IP-Routing-Table), from 10.3.1.7, 00:12:59 ago
The following output shows forwarding information in which the next hop for the route, 10.3.1.2, is the PE-3 router rather than the PE-2 router (which is the best path according to OSPF). The reason the OSPF route is not redistributed to BGP on the PE is because the other end of the sham-link already redistributed the route to BGP and there is no need for duplication. The OSPF sham-link is used only to influence intra-area path selection. When sending traffic to a particular destination, the PE router uses the MP-BGP forwarding information.
PE-1# show ip bgp vpnv4 all tag | begin 10.3.1.7
10.3.1.7/32 10.3.1.2
notag/38
PE-1# show tag-switching forwarding 10.3.1.2
Local Outgoing Prefix Bytes tag Outgoing Next Hop
tag tag or VC or Tunnel Id switched interface
31 42 10.3.1.2/32
0 PO3/0/0 point2point
PE-1# show ip cef vrf ospf 10.3.1.7
10.3.1.7/32, version 73, epoch 0, cached adjacency to POS3/0/0
0 packets, 0 bytes
tag information set
local tag: VPN-route-head
fast tag rewrite with PO3/0/0, point2point, tags imposed: {42 38
}
via 10.3.1.2
, 0 dependencies, recursive
next hop 10.1.1.17, POS3/0/0 via 10.3.1.2/32
valid cached adjacency
tag rewrite with PO3/0/0, point2point, tags imposed: {42 38}
If a prefix is learned across the sham-link and the path via the sham-link is selected as the best, the PE router does not generate an MP-BGP update for the prefix. It is not possible to route traffic from one sham-link over another sham-link.
In the following output, PE-2 shows how an MP-BGP update for the prefix is not generated. Although 10.3.1.7/32 has been learned via OSPF across the sham-link as shown in bold, no local generation of a route into BGP is performed. The only entry within the BGP table is the MP-BGP update received from PE-3 (the egress PE router for the 10.3.1.7/32 prefix).
PE-2# show ip route vrf ospf 10.3.1.7
Routing entry for 10.3.1.7/32
Known via "ospf 100
", distance 110, metric 12, type intra area
Redistributing via bgp 215
Last update from 10.3.1.2 00:00:10 ago
Routing Descriptor Blocks:
* 10.3.1.2 (Default-IP-Routing-Table), from 10.3.1.7, 00:00:10 ago
Route metric is 12, traffic share count is 1
PE-2# show ip bgp vpnv4 all 10.3.1.7
BGP routing table entry for 100:251:10.3.1.7/32, version 166
Paths: (1 available, best #1)
Not advertised to any peer
Local
10.3.1.2 (metric 30) from 10.3.1.2 (10.3.1.2)
Origin incomplete, metric 11, localpref 100, valid, internal,
best
Extended Community: RT:1:793 OSPF DOMAIN ID:0.0.0.100 OSPF
RT:1:2:0 OSPF 2
The PE router uses the information received from MP-BGP to set the ongoing label stack of incoming packets, and to decide to which egress PE router to label switch the packets.