IP SLA for dynamic failure of BGP routes

[thiyagarajankalaiselvan]

Hello All,

Can anybody help me to know is it possible to use IP SLA for dynamic failure if a specific route is not received from specific neighbor.

For example, In a dual homed scenario , if a router is not receiving route for 0.0.0.0 from ISP A and it should failover to ISP B.

Regards,

Thiyagu

[John Blakley]

Do you have a peering between both ISPs?

If so, do you get the same route from both ISPs?

Do you want to fail over completely to the other ISP or do you want to fail over for the single route?

HTH,

John

[thiyagarajankalaiselvan]

Hi ,

We do have peering between both the ISPs and we want to completely failover to the secondary ISP.

Regards.

Thiyagu

[John Blakley]

If you're wanting to prefer one isp over the other, you can use weights or local pref to do that. If the ISP were to completely go down, it would automatically fail over to the other isp for backup.

[thiyagarajankalaiselvan]

Thanks John for your reply. The CE is receiving a default route from PE. Even if there is any connectivity issue between PE and P or anywhere in upstream after PE, still PE is advertising a default route to CE so traffic from CE is reaching PE and it is getting dropped.

I'm trying to find an option for failure detect of connectivity issue between PE and P using IP SLA and failover of BGP from primary to secondary path.

Is it possible to use IP SLA with BGP?

Regards,

Thiyagu

[John Blakley]

You stated that the PE is still advertising a default route if there's a problem anywhere upstream. That rules out trying to monitor your 0.0.0.0/0 route because it will always be there unless the PE goes down. So, what will you ping on the other side? You could ping a DNS server like 8.8.8.8, but what happens if that DNS server has a real problem and nothing is wrong with your provider, or there's something else wrong along the path to the destination that's out of yours or your provider's control like another ISP? You'll fail over regardless if there's a real problem on your end. If you're okay with that, you can use sla to ping whatever you want to ping (maybe you have something that you own on the other side that's reliable).

The problem is that you are receiving a route from your ISP. Let's assume that route never leaves your table. The only way that you can get that route out of the routing table is for the provider to stop advertising it, you filtering it, or another route with a better AD is available. You can configure IP SLA to ping whatever you want, create a tracked object based on that, and then run a eem script based on the tracked object. The option to have a default 0/0 static route with a higher AD and tracked won't work. The route would be available when the track is up.

Another option would be to create two default routes - both with lower ADs than BGP (20). Then track the ISPA route:

ip sla 1

icmp-echo 8.8.8.8

freq 5

track 1 ip sla 1 reachability

ip sla schedule 1 start now life forever

ip route 0.0.0.0 0.0.0.0 <ISPA> 3 track 1

ip route 0.0.0.0 0.0.0.0 <ISPB> 4

What the above will do is prefer ispa if the track is up. If it goes down, the route to ispb will get put in the routing table. You would never use the default route that the provider is handing over because the bgp route is a higher ad.

The eem script is a lot more complicated than above, plus you'd have to have two scripts - one for when you see a problem and one for recovery.

The only other thing that I could recommend would be to look into PfR. It may not suit your needs though:

http://www.cisco.com/c/en/us/td/docs/ios-xml/ios/oer/configuration/12-4t/oer-12-4t-book/oer-setup-ntwrk.html

HTH,

John

*** Please rate all useful posts ***