Resource requests and limits

Kubernetes allows us to achieve high utilization of our cluster by scheduling multiple different workloads to a single pool of machines. Whenever we ask Kubernetes to schedule a pod, it needs to consider which node to place it on. The scheduler can make much better decisions about where to place a pod if we can give it some information about the resources that the pod will need; it then can calculate the current workload on each node and choose the node that fits the expected resource usage of our pod. We can optionally give Kubernetes this information with resource requests. Requests are considered at the time when a pod is scheduled to a node. Requests do not provide any limit to the amount of resources a pod may consume once it is running on a particular node, they just represent an accounting of the requests that we, the cluster operator, made...