I’d like to implement a fairly minimal infrastructure design that enables an Active-Active connection between two data centers (for the runtime resources at least, one of the orchestrator could be passive, if it is a viable option).
As I understand, for having 2 Orchestrator servers operating simultaneously, we need the High Availability Add-On installed on 3 different servers (1 master+2 secondary, but 1 master server is absolutely necessary).
If we would leave out the HAA with its 3 servers, but configure the load balancer to open the connections of the runtime resources within both two separate geolocations to the Orchestrator in the primary geolocation if it’s available, but redirect them to the other Orchestrator in the secondary geolocation if the primary is unavailable, would it cause any issues within the infrastructure?
The Orchestrators would connect to the SQL databases through an SQL Listener. We don’t necessarily need the two Orchestrators for balancing the load between them, but to ensure that one is definitely up and running, and can immediately take over operating our robots if the primary one fails.
If HAA is an absolute must in this case as well, would it be enough to install only 1 HAA server within the primary location?