topic Dual ISPs VPN failover across both in General Topics

Dual ISPs VPN failover across both

drewdown — Thu, 18 Aug 2022 18:59:42 GMT

Trying to provide some tunnel redundancy to some of our AWS environments. I have 2 ISPs both with an interface/static IPs on my HA PANs. ISP-A is my default with a default route to the internet pointing to its next hop.

ISP- A Eth1/8 9.9.9.9/24 ZONE-A

ISP-B Eth1/7 10.10.10.10/24 ZONE-B

Currently have all my VPN tunnels across ISP-A and want to bring up some tunnels across ISP-B. I created the IKE gateways and attached them to Eth1/7 ISP-B but in the logs I see the VPN traffic being dropped because its sending the traffic from Eth1/7 to Eth1/8 which is wrong, it should just go from Eth1/7 to the internet. I am missing something but not sure what.

Do both public interfaces need to be in the same ZONE? That is the only thing I can really see that maybe causing it. We have 2 static routes pointing out each ISP with the metric on the ISP-B higher than ISP-A so it prefers that route.

How do I force specific VPN tunnel traffic out of the secondary interface? I thought just trying the gateways to that interface would suffice but it doesn't appear to be the case.

Do I need static routes pointing to ISP-B for each tunnel endpoint at AWS?

Re: Dual ISPs VPN failover across both

OtakarKlier — Fri, 19 Aug 2022 18:34:10 GMT

Hello,

First the zone question. Basically its your preference, zones are there to provide logical separation so that security policies can/will be applied to the traffic.

Second the VPN tunnels: What I did was create the tunnels to AWS. Then I created Policy Based Forwarding rules to force the traffic down one tunnel and have the tunnel removed if it fails.

This way I dont have traffic dropped due to asymmetric return. AWS is all policy based routing so it gets goofy under the hood. This is why I chose the PBF rules instead of dynamic routing etc.

Hope that makes sense.

Re: Dual ISPs VPN failover across both

drewdown — Mon, 22 Aug 2022 14:22:15 GMT

I don't mess with PBF at all anymore as it was more work than it's worth. For redundancy to the internet we utilize 2 static routes (1 with a higher cost) and path monitoring. That eliminates the need for PBF and having to make sure all the right subnets are configured and all that noise.

As far as VPN failover I got it working by simply adding static routes pointing to the secondary AWS tunnel IPs via the secondary circuit and run BGP across all of it. Works great.