Support HA (High Availability)

What would it take for the Hub, and the Proxy to support HA?

i.e. to have more than 1 pod (multi-head), ideally 3 or more so in case of some AWS EC2 instances going down the service survives?