Set Kubernetes CPU and memory requests and limits and determine QoS class.
Requests drive scheduling and eviction priority while limits cap actual usage. Guaranteed QoS requires requests to equal limits, which protects the pod from eviction but wastes headroom. CPU limits cause throttling rather than failure while memory limits cause hard termination, so the two should be set with different degrees of caution.
Container Resource
Requests sit just above observed use; limits allow for peaks unless QoS is Guaranteed
Requests sit just above observed use; limits allow for peaks unless QoS is Guaranteed Requests drive scheduling and eviction priority while limits cap actual usage. Guaranteed QoS requires requests to equal limits, which protects the pod from eviction but wastes headroom.
CPU limits cause throttling rather than failure while memory limits cause hard termination, so the two should be set with different degrees of caution.
This calculator takes 5 inputs: Observed CPU use, Observed memory use, Peak to average ratio, Replicas, QoS target. The pre-filled defaults are a realistic starting point — replace them with figures from your own environment for a result you can act on.