EIGRP - How can I prefer an Externally learnt route to an Internally Learnt Route

[ccannon88567]

I'm hoping somebody can advise me on an issue I have?

We have two data centre's connected to an MPLS cloud. Routes are re-distributed from BGP into EIGRP at both data centres. The routes are learnt as external so get a 170 AD in EIGRP for each DC which is then passed around the DC's.

However the routes are then learnt via a secondary direct LES link that exists directly between the two data centres which are classed as internal routes.

How can I force or manipulate EIGRP to prefer the externally learned routes so that traffic between the DC's takes the MPLS and uses the LES only as a backup?

Maybe set the AD lower than 90 for externally learnt routes? Would the AD propogate to other routers?

Many thanks

[Adam Casella]

Hey,

Without knowing your setup better, it sounds like MPLS VPN might be a good option to use here.   The BGP device would be setup as PE devices and you could then pass these routes from one location to the other as local eigrp routes.    The basic topology is the following:

(site2)CE----(EIGRP)PE(BGP)-----MPLS-------(BGP)PE(EIGRP)-----CE(site1)

These types of setups are not trival and require indepth knowledge of VRF's, MPLS VPN's (route targets and route descriptors) as well as MP-BGP.   However the routes passed over the VPN will look like intra-AS EIGRP routes and cost can then be specified.

You can review more about this via the following link:

http://www.cisco.com/en/US/docs/ios/12_2t/12_2t15/feature/guide/fteipece.html#wp1027258

Admin distance is locally significant and is not passed.  This would not be feasible if you have a large EIGRP AS and may have undesired effects if you are redistributing EIGRP routes into another routing protocol (routing loops, etc).

Thanks,

Adam

[ccannon88567]

Hi Adam,

I'm hoping that there may be a simpler solution than setting up VRF's. We have limited access to the CPE and no access to the PE, plus the SP may already be using VRF's of some type to keep our routes private.

We redistribute on the CPE and this runs both EIGRP and BGP. Because the router is reidtributing the routes are AD 170. Routers downstream within both data centres can see each other via a metro ethernet link. Because this is internally learnt they are communicating over this link and I need them to prefer the MPLS.

Thank you for your help.

[mikegrous]

Semi differnt scenario but same logic...This might help...  1 Hub router having 2 Spokes

Route table from Hub....

     1.0.0.0/24 is subnetted, 1 subnets
D       1.1.1.0 [90/156160] via 10.2.2.3, 00:00:03, FastEthernet0/1
                [90/156160] via 10.1.1.1, 00:00:03, FastEthernet0/0
     2.0.0.0/24 is subnetted, 1 subnets
D       2.2.2.0 [90/156160] via 10.2.2.3, 00:00:03, FastEthernet0/1
                [90/156160] via 10.1.1.1, 00:00:03, FastEthernet0/0
     10.0.0.0/24 is subnetted, 2 subnets
C       10.2.2.0 is directly connected, FastEthernet0/1
C       10.1.1.0 is directly connected, FastEthernet0/0

(Added commands)

router eigrp 1

  distance 69 10.1.1.1 0.0.0.0

*Mar  1 00:06:26.591: %DUAL-5-NBRCHANGE: IP-EIGRP(0) 1: Neighbor 10.1.1.1 (FastEthernet0/0) is down: route configuration changed
*Mar  1 00:06:26.603: %DUAL-5-NBRCHANGE: IP-EIGRP(0) 1: Neighbor 10.2.2.3 (FastEthernet0/1) is down: route configuration changed
*Mar  1 00:06:28.191: %DUAL-5-NBRCHANGE: IP-EIGRP(0) 1: Neighbor 10.2.2.3 (FastEthernet0/1) is up: new adjacency
*Mar  1 00:06:31.015: %DUAL-5-NBRCHANGE: IP-EIGRP(0) 1: Neighbor 10.1.1.1 (FastEthernet0/0) is up: new adjacency
NOTE EIGRP FLAPS

sho ip route from hub

     1.0.0.0/24 is subnetted, 1 subnets
D       1.1.1.0 [69/156160] via 10.1.1.1, 00:00:19, FastEthernet0/0
     2.0.0.0/24 is subnetted, 1 subnets
D       2.2.2.0 [69/156160] via 10.1.1.1, 00:00:19, FastEthernet0/0
     10.0.0.0/24 is subnetted, 2 subnets
C       10.2.2.0 is directly connected, FastEthernet0/1
C       10.1.1.0 is directly connected, FastEthernet0/0

admin distance changed only for routes sourced from 10.1.1.1. Hope it helps.