Each Service Fabric cluster node runs an agent that collects and sends load reports to a centralized resource balancer service. The resource balancer service is responsible for generating placement recommendations based on the load on the nodes and the placement requirements. The agent also sends failure reports and other events to a failover manager service. In a event of a change to the available node count, such as when a node fails or when a node is added to the cluster, the failover manager communicates with the resource balancer to get a placement recommendation. Once the failover manager receives a recommendation from the resource balancer service, it places a new service replica on the recommended node.