In which use case Multiple renderer is used in Dispatcher - aem

Question

I am trying to understand when it make sense to use multiple renderer in dispatcher.any configuration. For example I have this setup.
On Dispatcher-1 server:
/renders {
/0001 {
/hostname "https://publish-1.com"
/port "8443"
}
/0002 {
/hostname "https://publish-2.com"
/port "8443"
}
}
On Dispatcher-2 server:
/renders {
/0001 {
/hostname "https://publish-1.com"
/port "8443"
}
/0002 {
/hostname "https://publish-2.com"
/port "8443"
}
}
Now as per my understanding dispatcher will forward the traffic to either publish-1 or publish-2 server depending on the time taken by the respective publish server.
When I look into this page then I see there is no configuration which makes it possible to direct the traffic from Dispatcher-1 to Publish-2. It is always Dispatcher-1 to Publish-1 or Dispatcher-2 to Publish-2. If Publish-2 goes down then the request landing on Dispatcher-2 will not be processed.
Could you please clarify what is the use of only one-to-one setup and when it make sense to have one-to-many setup?
Thanks for your help!

Answer 1

From your link
With stateful authentication, the persisted authentication state will only be available on the instance where the user is first authenticated.
This is the key. Since Publishers B & C don’t know about the authenticated request, and sessions are not shared between the application servers, all subsequent requests need to be “sticky” to Publisher A. This is maintained by the load balancer, not the dispatcher.
As a result, the dispatcher in front of Publisher A is 1:1 mapped to a single node rather than many.