I am attempting to connect two AWS regions. AWS's documentation suggests starting up an instance on both sides to run software IPSec (OpenSWAN or StrongSWAN), giving both instances an elastic IP and using that as a tunnel. That's all well and good, but now we have a single point of failure on both sides.
AWS released a VPC Connectivity Guide in July that details the possible options for connecting two VPCs together. I've deep-linked that PDF to page 15, where it explains the options.
Every option it lists has significant downsides, since in most cases all traffic goes through a single (crashable) instance. So far it seems that the best options is to use a software tunnel on one side, and a Virtual Private Gateway on the other. They say, at least you get redundancy on one side. That still seems suboptimal.
Is there a way to connect two Virtual Private Gateways together so I get the best of both worlds - Amazon-managed redundancy and the simplicity of keeping all of this inside my VPC configuration? Or are VGWs essentially client-only?