Cisco Support Community
cancel
Showing results for 
Search instead for 
Did you mean: 
Announcements

Welcome to Cisco Support Community. We would love to have your feedback.

For an introduction to the new site, click here. And see here for current known issues.

New Member

BGP and OSPF Load Balance

Dear All,

I have an odd case please see the diagram link below, i need reach from R-C01 (192.168.253.0/24) to 192.168.90.0/24 load balance or load  sharing via R-BH-01 and RR01 is it posible ? how to  configure it?

http://imageupload.org/di-6128999679510.jpg

1 ACCEPTED SOLUTION

Accepted Solutions
Cisco Employee

Re: BGP and OSPF Load Balance

Hello,

From your exhibit, I assume that R-C01 knows about the network 192.168.90.0/24 via OSPF only, and that the route is learned as external (redistributed). You have not indicated whether the route is advertised as External Type-1 or External Type-2.

In general, the OSPF can perform only equal-cost load balancing. That implicates that the R-C01 must consider both routes to the network 192.168.90.0/24 of equal cost, only then it will perform load balancing. In turn, that means:

  1. If the network is advertised as External Type-1 then the total metric of both routes must be equal, i.e.: the sum of link costs R-C01:RR01+RR01:RR02+RR02:R-C02+R-C02:R-BH-02+seed metric must be equal to the sum of link costs R-C01:R-BH-01+seed metric
  2. If the network is advertised as External Type-2 then the seed metric must be identical on both redistributing routers, and the cost to reach both ASBRs must be identical, i.e. R-C01:RR01+RR01:RR02+RR02:R-C02+R-C02:R-BH-02 = R-C01:R-BH-01

Obviously, you would need to manually modify the link costs to get the desired effect.

As a general rule, a routing protocol is responsible for populating the routing table with accurate and complete information about best paths to any reachable destination. However, what you want to do is basically traffic engineering, and that is out of scope of routing protocol competences. We can force a routing protocol to perform load balancing for a certain set of destinations but for more general deployments, we have to resort to specific traffic engineering mechanisms such as MPLS-TE.

Best regards,

Peter

2 REPLIES
Cisco Employee

Re: BGP and OSPF Load Balance

Hello,

From your exhibit, I assume that R-C01 knows about the network 192.168.90.0/24 via OSPF only, and that the route is learned as external (redistributed). You have not indicated whether the route is advertised as External Type-1 or External Type-2.

In general, the OSPF can perform only equal-cost load balancing. That implicates that the R-C01 must consider both routes to the network 192.168.90.0/24 of equal cost, only then it will perform load balancing. In turn, that means:

  1. If the network is advertised as External Type-1 then the total metric of both routes must be equal, i.e.: the sum of link costs R-C01:RR01+RR01:RR02+RR02:R-C02+R-C02:R-BH-02+seed metric must be equal to the sum of link costs R-C01:R-BH-01+seed metric
  2. If the network is advertised as External Type-2 then the seed metric must be identical on both redistributing routers, and the cost to reach both ASBRs must be identical, i.e. R-C01:RR01+RR01:RR02+RR02:R-C02+R-C02:R-BH-02 = R-C01:R-BH-01

Obviously, you would need to manually modify the link costs to get the desired effect.

As a general rule, a routing protocol is responsible for populating the routing table with accurate and complete information about best paths to any reachable destination. However, what you want to do is basically traffic engineering, and that is out of scope of routing protocol competences. We can force a routing protocol to perform load balancing for a certain set of destinations but for more general deployments, we have to resort to specific traffic engineering mechanisms such as MPLS-TE.

Best regards,

Peter

New Member

Re: BGP and OSPF Load Balance

Hi Peter,

     Many thanks for your reponse, yes it works i did manipulate the ospf cost and med for BGP, i'm just afraid after i implement in production it could caused asymetric routing or somthing.

R-BH-01
!
Router ospf 600
redistribute bgp 600 metric 299 metric-type 1 subnets
!
router bgp 600
redistribute ospf 600 metric 100 match internal external 1 external 2
!


R-BH-02
!
router ospf 600
redistribute bgp 600 metric 97 metric-type 1 subnets
!
router bgp 600
redistribute ospf 600 metric 100 match internal external 1 external 2
neighbor loadsharing route-map med_600 out
!
access-list 66 permit 192.168.253.0 0.0.0.255
!
route-map med_600 permit 10
match ip address 66
set metric 90
!
route-map med_600 permit 100
set metric 200
!

507
Views
0
Helpful
2
Replies
CreatePlease login to create content