What it is
Kubernetes networking allocates a pod CIDR range per node, and the CNI plugin you choose (Calico, Cilium, AWS VPC CNI, and others) affects how that range is subdivided and how many pods a node can host.
When to use it
Platform engineers use this before standing up a new cluster, to avoid exhausting the pod CIDR range as the cluster scales or colliding with existing VPC subnets.
Gotchas & limits
Pod CIDR sizing decisions are hard to change after a cluster is running; undersizing max-pods-per-node early on is one of the most common causes of a painful re-platforming later.