On-prem group is a logical grouping of one or more on-prem agents into a common connection pool. The workload is distributed evenly to all agents in an on-prem group. A recipe that uses an on-prem group with more than one active agent achieves High availability. This has a few advantages.
High Availability is the ability of a system to be continuously operational for a long period of time. For any system to be highly available, every part of a system needs to be present and functional for the whole to be operational. This is typically achieved through Fault tolerance and Load balancing.
Fault tolerance allows a system to continue operating properly in the event of the failure of some components. In on-prem groups, this is achieved by adding redundancy to the on-prem agents. Each of these agents acts as a backup or fail-safe, in the event that one or more server hosting them goes offline.
Load balancing is the distribution of work across multiple resources. In this case, we are distributing requests across multiple on-prem agents in the same group. This avoids overloading a single server, improves responsiveness and overall performance of a recipe using on-prem connections.