1 of 1 people found this helpful
For an external load balancer, it can more than likely round robin incoming requests to any nodes within the Tableau cluster that are running the "Gateway" service and can be setup that if it detects the Gateway service is down on a node it can stop directing requests to that node until the Gateway service becomes available again.
There is internal load balancing that goes on based on nodes that are running the Gateway service. As an example, let's say you have a 5 node cluster and I'm going to use a stripped down example:
Let's assume all 5 nodes are running the Gateway service but I've called out an additional service for each of the nodes
A - VizQL Server
B - VizQL Server
C - VizQL Server
D - Backgrounder
The nodes themselves know what services are available on the other nodes. Let's say the Gateway on node "E" gets a request to render a visualization for an end user. It knows nodes, A, B, and C are running the VizQL Servers but will start with the first A. The next time the Gateway on node "E" gets the same type of request it will route to node "B". It basically does a "round robin" of those nodes that it knows run the service.