Redundancy and Default routes - EIGRP

[Jules Sanchez]

Hello,

We have 2 sites all running EIGRP.

Both sites advertise all our networks via BGP, 1/2 are prepended, the other half are not, basic redundancy.

So if we lose 1 Internet link, the other site picks up the slack.

The 2 Cores are interconnected via a point to point.

Here's my crude diagram

          INTERNET

           /          \

         /             \

EDGE 1          EDGE 2

     |                      |

     |                      |

CORE 1 -------- CORE 2

     |                      |

(other routers)        (other routers)

At this point, all these routers have a default route.

EDGE 1

ip route 0.0.0.0 0.0.0.0 INTERNET (provider 1)

EDGE 2

ip route 0.0.0.0 0.0.0.0 INTERNET (provider 2)

CORE 1

ip route 0.0.0.0 0.0.0.0 EDGE 1

CORE 2

ip route 0.0.0.0 0.0.0.0 EDGE 2

Which just doesn't work in the following scenarios.

We lose Internet on EDGE 1.

All BGP routes are now inbound to EDGE 2, all networks can be reached via EIGRP (CORE 1 and CORE 2), but the default route on CORE 1 is EDGE 1, which has no Internet link !

I have redistribute static (into EIGRP) on all routers, and removed the default route from CORE 1.

But it picked up the default route from another router down the chain, not the EDGE 1 router.

What would be the best way to handle this ?

Do I remove all the default routes from every router and just keep them on the EDGE routers ?

Even then, how would the CORE know which to choose ?

Can anyone recommend a better way ?

Thanks for your time

-

Jules.

[Lei Tian · Accepted solution]

Hi Jules,

If you have available ports between EDGE routers, you can run iBGP peer between the edge routers, and only advertise default to the cores. That way the EDGE routers will be the points to make the route decision on which SP to use.

If you dont have available ports between EDGE routers, you can use reliable static (static route with track option) on COREs; it can track the reachability to the SP. If the EDGE 1 lost Internet, core 1 will remove the static route and it will follow the default route learned from core 2 to reach EDGE 2. 

HTH,

Lei Tian

[Jules Sanchez]

Hi Lei Tan,

That sounds like what I need, static routes with tracking.

On each of the CORE routers do I ping the next hop ?

CORE 1

currently has

ip route 0.0.0.0 0.0.0.0 EDGE 1

Do I remove this route and put in the tracking route ?

ip sla 1 etc...

icmp-echo etc..

Can you give me an example ?

Thanks

[Lei Tian · Accepted solution]

Hi Jules,

Yes, you got the idea. Just replace your current static route with a reliable static route. Here is an config example

'Configuring Reliable Static Routing Backup Using Object Tracking: Static Routing Examples'

http://www.cisco.com/en/US/docs/ios/12_3/12_3x/12_3xe/feature/guide/dbackupx.html#wp1066247

Regards,

Lei Tian

[Jules Sanchez]

Ok so let's say the following;

This is CORE 1.

2 default routes.

ip route 0.0.0.0 0.0.0.0 100.1.1.242 track 123 
(IP address is NEXT HOP to Provider 1)

ip route 0.0.0.0 0.0.0.0 200.2.2.125 254
(IP address is CORE 2).

Provider 1 goes down, Core can no longer ping 10.1.1.242
and drops the 1st static route and uses the 2nd static route.

It now goes through CORE 2 and Provider 2 Internet, 
however now it (CORE 1) can see 100.1.1.242 because 
it has Internet access again through Provider 2.

Won't this screw things up ?

[Lei Tian · Accepted solution]

Hi Jules,

If the sp's addresses are routable, and interface remains up/up when lose the link; that is possible. One option is, you can config a static route on core, for example ip route 100.1.1.242 edge1 on core 1, and ip route 200.2.2.125 edge2 on core 2. That will force core 1 to use edge1 to check reachability for provider 1.

HTH,

Lei Tian

[Jules Sanchez]

You're a genius

Thanks so much for your help.

[Utair Corporation]

Just a thought, i'm not goot at BGP.

Couldn't ISPs provide you with default route via BGP? Which you would redistribute into EIGRP.

I case of troubles your network would switch to backup link, not tracking needed.