I currently have the following setup:
1) I am provided a /30 peering address with my ISP. Let's say this is 1.2.3.5/30 and my next hop is 1.2.3.6/30
2) I have an interface on a solo checkpoint assigned 1.2.3.5/30 respectively (say eth2)
3) I advertise a /27 BGP block to this next hop. For example: 71.82.42.224/27. Any traffic that has arrived here is NAT'ed to some internal destination.
4) I have several IPSEC VPNs (Domain Based) with 3rd parties that terminate on the PEERING address (1.2.3.5/30 eth2)
All is well in this case.
ISP changes are now requiring me to change the peering address, but I can keep the /27 block.
Therefore, I now want to change point #4 above. ie: terminate VPNs on one of my BGP addresses.
So far, my thoughts have been to create a new interface (in the same WAN VLAN? Loopback?) and assign it an IP in the BGP block so that the ipsec daemon can terminate there. (Let's say 71.82.42.225 eth3) But then what?
This new interface appears in the selection box in gateway properties --> ipsec vpn --> link selection - but this would break all EXISTING VPNs until the 3rd party changes their side? These are customers (about 20 of them), taking them down isn't an option nor is trying to align 20 customers on the same day to change their side.
Is there a way I can gradually migrate each of my VPNs? ie: keep unmigrated customers IPSEC VPNs terminated on 1.2.3.5 while I migrate other customers, 1 by 1, to use the BGP IP 71.82.42.225?
Is it as simple as NAT? (Right now, when the CP initiates IPSEC it always wants to use its main peering IP)
Policy based routing? (Although no matter what, the ISP routes traffic to the BGP block to my peer address?)
If someone has any other clever ideas, I'm all ears.
Thanks in advance.