HP FlexNetwork 10500 Series Configuration Manual page 206

Routing loop avoidance
Figure 57 Network diagram for routing loop avoidance
VPN 1
OSPF Area 0
Site 1
As shown in
from MP-BGP to Site 1 through OSPF, the routes might be received by the other PE. This results in a
routing loop.
OSPF VPN extension uses the following tags to avoid routing loops:
•
DN bit (for Type 3 LSAs)—When a PE redistributes BGP routes into OSPF and creates Type 3
LSAs, it sets the DN bit for the LSAs. When receiving the Type 3 LSAs advertised by CE 11, the
other PE ignores the LSAs whose DN bit is set to avoid routing loops.
•
Route tag (for Type 5 or 7 LSAs)—The two PEs use the same route tag. When a PE
redistributes BGP routes into OSPF and creates Type 5 or 7 LSAs, it adds the route tag to the
LSAs. When receiving the Type 5 or 7 LSAs advertised by CE 11, the other PE compares the
route tag in the LSAs against the local route tag. If they are the same, the PE ignores the LSAs
to avoid routing loops.
OSPF sham link
As shown in
•
A route over MPLS backbone—It is an inter-area route if PE 1 and PE 2 have the same
domain ID, or is an external route if PE 1 and PE 2 are configured with no domain ID or with
different domain IDs.
•
A direct route between CEs—It is an intra-area route that is called a backdoor link.
VPN traffic is always forwarded through the backdoor link because it has a higher priority than the
inter-area route. To forward VPN traffic over the inter-area route, you can establish a sham link
between the two PEs to change the inter-area route to an intra-area route.
CE 11
Figure 57, Site 1 is connected to two PEs. When a PE advertises VPN routes learned
Figure 58, two routes exist between Site 1 and Site 2 of VPN 1:
PE 1
MPLS backbone
PE 2
197
VPN 1
OSPF Area 1
Site 2
CE 21
PE 3
CE 22
VPN 1
OSPF Area 2
Site 3