Is there any reason you cannot create a 2-node failover cluster for Alteryx Server?
In this example you would have the Controller, Gallery & Workers on both nodes (MongoDB being run externally on Mongo Atlas) with the MS Cluster Failover Roles configured to manage the Active Controller (following the HA instructions described here)