BGP connection to 2 ISP and Internet outage

[mahesh18]

Hi Everyone,

We have 2 ASR  routers and each one is having connection to unique ISP.

I have attached the diagram and BGP config from both ASR1 and ASR2.

ASR1 and ASR2  have bgp nei between them using 192.41.x.x subnet and they learn all the BGP routes

from the internet.

Tracert from PC

tracert from PC

byte packets
1 198.160.191.2 (198.160.191.2) 0.512 ms 0.478 ms 0.558 ms-----------------------------------------AS2
2 192.41.148.211 (192.41.148.211) 0.549 ms 0.536 ms 0.527 ms-----------------------------------------------ASR1
3 ra2so-ge4-2-24.cg.bigpipeinc.com (64.141.118.153) 1.554 ms 1.619 ms 1.698 ms

Recently our ISP 1 had outage and we had internet outage--

1> ASR1 shows BGP connection up from 1 year.

2>ASR2 shows BGP Connection to ISP as uptime from 2 hours  that shows it went down.

Need to know why ASR1  shows BGP nei to ISP as always up?

What config we can do on ASR1 that can tell us if connection to ISP1 is down then it can failover to ISP2?

Also as ASR1 and ASR2 have BGP peering between them how do they learn whole internet routing table from the internet?

Regards

Mahesh

[chrihussey · Accepted solution]

Hello,

Your first question:

Need to know why ASR1  shows BGP nei to ISP as always up?

ANS. Just because ISP1 experienced an outage, does not mean that the actual link or peering between them and ASR1 failed. In that instance, the trace route would succeed and the peering would be maintained.

Second question:

Also as ASR1 and ASR2 have BGP peering between them how do they learn whole internet routing table from the internet?

ANS. If both ASRs receive full routing tables from their respective ISPs, they exchange these routes with each other via IBGP. Depending on how things are configured, they then determine which path (ISP) is the preferred route.

As for the outage, with full Internet routing tables being exchanged it could take some time for things to re-calculate, minutes or more, not seconds. Also, depending on the nature of the problem, routes could have been coming and going further exacerbating things. Just a guess though. Short of that, it may be a good idea to check your advertisements to ISP2 and validate that they are correct and that the ISP is advertising them correctly to the greater internet.

Hope this helps

[mahesh18]

Hi,

Thanks for the reply.

Can you please provide me more info on this 

ANS. If both ASRs receive full routing tables from their respective ISPs, they exchange these routes with each other via IBGP. Depending on how things are configured, they then determine which path (ISP) is the preferred route

[chrihussey · Accepted solution]

So for an example, if ASR1 receives a network route from ISP1, and ASR2 receives the same network from ISP2, but the route from ISP1 is fewer AS hops. ASR2 learns this via IBGP and prefers the route via ASR1. Similarly, routing could be influenced in many ways, such as weighting, local preference, metrics etc. 

In instances where a path is equal for each ASR to their ISP, then it will be the preferred path in that router.

Regardless, if ASR1 & 2 are IBGP peering and exchanging learned routes from their respective ISPs, between the two they determine the preferred path.

Hope that makes sense and helps.

[mahesh18]

Many thanks for your Answers.

Best Regards

MAhesh

[jmperlewitz]

You can create an EEM script that uses IP-SLA to track ICMP communication to an IP on the internet.  If it loses connectivity (but the BGP peer stays up), it will shut down the BGP peer (or the interface).