Kubernetes resource reservation per namespace

Question

How can I reserved resources to a namespace?

Example: I want to guarantee that one namespace cannot allocate all the resources on the cluster.

Example: Having namespace A and B and a Cluster that can have 100 pods. How can I make sure that the pod A can schedule at least 10 pods, even if we have 200 users trying to create pods on namespace B. (Typical resource segregation)

I would expect something like in Yarn where I can say leave 10% of the cluster resources to queue A.

Answer 1

Namespace resource quota won't do this for you. Try to explore: http://yunikorn.apache.org/ . More specifically the min/max capacity model: http://yunikorn.apache.org/docs/next/get_started/core_features#hierarchy-resource-queues .

Answer 2

Resource quotas are a tool for administrators to address this. Users create resources (pods, services, etc.) in the namespace, and the quota system tracks usage to ensure it does not exceed hard resource limits defined in a ResourceQuota