What it is
Kubernetes components never call each other: they all talk to the apiserver and react to watch events. The scheduler takes each unscheduled pod, filters out nodes that cannot run it, scores the rest, and binds the pod to the winner.
When to use it
Use it to learn what really happens between kubectl apply and a Running pod, to understand why a pod is Pending, or to compare scheduling strategies, taints, affinity, topology spread and preemption, and to rehearse Service endpoints, autoscaling, PodDisruptionBudgets and volume binding before touching a real cluster.
Gotchas & limits
This is an educational model of Kubernetes v1.37 behaviour. Scheduling, EndpointSlices, PodDisruptionBudgets, volume binding and the HPA were compared with recordings of the real components; CNI, DNS, LoadBalancers, probes and real-world timings are simplified, so it does not replace testing on a real cluster.