Load sharing configurations

Load sharing distributes the system's processing load between two nodes. The two nodes can also be configured to backup each other.

The primary benefit of load sharing is to reduce failover time: only one half of the objects must fail over.

A system with redundant AppEngines can host Active and Standby Engines of different redundant pairs on the same node. This configuration enables efficient use of system resources while providing high availability for both nodes.

However, configuration of redundant engines on the same node requires thorough evaluation of the following critical (Performance Monitor) counters: CPU and memory use on each node.

Ensure each AOS node runs at a maximum of 25-30% CPU load. This CPU load represents the resources used by each AOS in steady condition considering Active and Standby engines running in each node as well as any other applications. When a failover occurs one of the AOS nodes will host all the Active engines in the pair. Hosting implies a CPU load increase to a total of 50-60%.

It is important to note that although the computers hosting the redundant AppEngines do not have to be identical, the total load of the application must not exceed 60% of CPU usage of the smallest computer.