cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
9323
Views
13
Helpful
29
Replies

Prefer BGP routes over EIGRP

aijaz802
Level 1
Level 1

Hi,

I have two links from site A to Site B one link 2 MB is directly connected and other one is of 4 MB thru the service provider. I'm using EIGRP b/w Sites.

Using EBGP b/w router A-SP & SP-B for the service provider network. On router C I'm seeing the Router B network routes which are redistributed into EBGP. Everything is fine so far.

On site A Router A EIGRP topology contains two routes one learned thru EIGRP (Rtr B) n other one thru EIGRP-EBGP (Rtr C) and Router A injecting routes learned thru EIGRP internal) even though the BGP learned routes are having less metric, I think its due to external routes.

I want the Site A to use the path A-C-B, instead of A-D-B, irrespective of route type internal/external/etc....and backup path as A-D-B.

Any suggestions please. I have attached the sample image.

Regards,

29 Replies 29

Sorry, which one will work fine?

I was referring to the following statement:

ip summary-address eigrp 100 10.208.0.0 255.240.0.0

Regards,

Regards,
Harold Ritter, CCIE #4168 (EI, SP)

Ok thanks.

I will be applying it tomorrow.

I have applied the statements and everything is working as desired.

There is just one small issue. It takes 3 1/2 mins to failover to the backup link. Is there anyway to to reduce this time?

Thanks for your help.

This sounds like the maximum it would take for BGP to declare the session down. You could definitely reduce the BGP hold timers to get a better convergence time going from the BGP to EIGRP path.

Just out of curiosity, how did you test the BGP failure?

Regards,

Regards,
Harold Ritter, CCIE #4168 (EI, SP)

I shut the interface down on the remote site and basically timed how long it would take to reach the site again.

Convergence from EIGRP to BGP was quicker and without any ICMP drops.

You should modify the BGP keepalive and hold timers to reduce the delay in BGP to EIGRP convergence.

Regards,

Regards,
Harold Ritter, CCIE #4168 (EI, SP)

What would be a suitable setting for the keepalive, say if I wanted to reduce the hold timer to 60 sec?

Would this be a stable setting?

You could easily go lower than 60 seconds without jeopardizing the stability of the network. What is your convergement requirement? Would 30 seconds be enough? I have seen many networks running with values lower than that without any issues.

Regards,

Regards,
Harold Ritter, CCIE #4168 (EI, SP)

30 seconds is good.

Would be good if was even lower.

Can this change be done for one site or does it need to be done on all BGP sites?

You could probably go lower than 30 seconds but be careful.

If you want the convergence time to be the same for all sites then you need to change the default for all sites.

Regards,

Regards,
Harold Ritter, CCIE #4168 (EI, SP)

I changed the keepalive to 10 sec and the hold timer to 30.

I tested it by bringing down the link, but it took around 2 1/2 for the rerouting to take place.

Any ideas why this would happen with the given values?

BGP neighbor is 62.6.17.17, remote AS 12641, external link

Description: VPNv4 to PE2-DXB-ME

BGP version 4, remote router ID 62.6.19.199

BGP state = Established, up for 09:19:13

Last read 00:00:03, last write 00:00:02, hold time is 30, keepalive interval is 10 seconds

Configured hold time is 30,keepalive interval is 10 seconds Minimum holdtime from neighbor is 0 seconds

Neighbor capabilities:

should I use the 'neighbor X.X.X.X fall-over' in the BGP process for faster convergence?

ip summary-address eigrp 100 10.208.0.0 255.240.0.0