While load balancing is a key part of managing a cluster, this is not what a cluster resource manager does. A traditional load balancer can be of two types, a network load balancer (NLB) or an application load balancer (ALB). The primary job of a load balancer is to make sure that all of the services hosted receive a similar amount of work. Some load balancers are also capable of ensuring session stickiness and some are even capable or optimizing the request routing-based on the turnaround time or current machine load.

While some of these strategies are efficient and best suited for a monolithic or tiered architecture, they lack the agility to handle faults and upgrades. A more responsive, integrated solution is required to handle hyperscale deployments of Microservices. The Service Fabric cluster resource manager is not a network load balancer. While a traditional load balancer distributes the traffic to where services are hosted, the Service Fabric cluster resource manager moves services to where there is room for them, or to where they make sense based on other conditions. These conditions can be influenced by the resource utilization on a node, a fault, an upgrade request, or so on.

For instance, the cluster resource manager can move services from a node which is busy to a node which is underutilized based on the CPU and memory use. It can also move a service away from a node which is in the queue for an upgrade.