This option is recommended for medium deployments of Robots (more than 250 Unattended Robots or 2,500 Attended Robots) and environments where High Availability is a top priority. It offers both a better performance and failure resistance, as multiple Orchestrator nodes are available - when one fails, the others pick up the load. Horizontal scalability is also available, as it is possible to add another node if your Robot needs grow.

A downside is that this model is resource intensive, seeing that you need additional components such as Redis and a network load balancer.

Please also take into consideration the hardware requirements here and here.