I really don't think it will work like that. The polling engines have the specific nodes that they poll. If one goes down, the other does not take over. If you need redundancy, the failover engine is needed. The failover engine takes over all of the down server's tasks. You will need a failover engine for each poller. If you have multiple primary products (NPM, NCM, or APM) on your server, you will need a license for each primary product, per server. It kinda sucks that you can't have the pollers do the failover too.