The topology for the egress of MAN of a carrier:
GSR-1 ― NE80E-1 ― S8016-1
| | \/ |
| | /\ |
GSR-2 ― NE80E-2 ― S8016-2
Two NE80Es serve as the egress router of a MAN, connecting to both GSR-1 of J company and GSR-2 of C company on provincial backbone network; they for a ring network, and ISIS is enabled as IGP; two NE80Es set up IBGP neighbor with GSR-1 and GSR-2 respectively, advertising the route of MAN to provincial backbone network through BGP;
Each of two NE80E is connected to both S8016 as convergence equipment in MAN, realizing dual-homing connection. MAN uses OSPF as IGP, and two NE80Es advertise the default route to lead the uplink traffic to S8016. The OSPF cost of links from S8016 to two NE80Es equal, and generally, two pieces of default OSPF routes with equal cost to NE80E-1 and NE80E-2 will be generated.
One day, the users under MAN reported that their access to network is very slow. After checkup, it is found that the default route generated by BGP at NE80E-1 to GSR-1 changes to OSPF default route to NE80E-2; however, there is only one default route at S8016 at MAN to NE80E-2. The traffic from MAN is centralized to the link of S8016-NE80E-2 - GSR-2, resulting in congestion of link bandwidth.
Change the priority of BGP route of NE80E as to ensure that the priority of BGP is always higher than that of OSPF, avoiding the problem.
For example, in BGP view, execute the command:
preference 120 120 255
Change the priority of the route from BGP learned by its peer to 120, higher than that of OSPF_ASE route.
The change to the default route of S8016 is caused by that the default route at NE80E-1 is generated by OSPF. During the configuration for advertisement of OSPF default route, if the default route of the router is generated by the OSPF of the same progress, the router will not flood the default route to ohter ospf peers.
What causes the default route at NE80E-1 to change?
1. For OSPF, if the LSA for the default route is created at the local, the local router will not calculate the LSA representing for the default route received from other routers.
2. For the case, BGP route is set up first, and one default LSA is created. After the definition of default route advertise for OSPF, it will not calculate the default LSA from NE80E-2.
3. For uncertain reasons, the default route of BGP disappear, and NE80E-1 deletes the default LSA created by itself, but calculate the default LSA received from NE80E-2.
4. Once the BGP route is recovered, it will not take effect because the priority of OSPF route is higher than that of BGP (OSPF defaults ASE route to 150, and BGP to 255), and NE80E-1 cannot create the default LSA.
In planning for routes of MAN, the changes to default route should be considered. If multiple routing prototocols at network could generate default routes, the priority of routing protocol ranks the first in consideration.