I'm trying to calculate dominant shares of resources in Kubernetes based on what Dominant Resource Fairness(DRF) does. How can I implement it in Kubernetes?
There are no specific set patterns or guidelines for DRF in Kubernetes yet. The closest that you can get to it is by trying to manage compute resources in containers together with what the kube-scheduler does. You can also start if you fetch all the pod requests and node capacities using kubectl.
As a matter of fact, the Kubernetes Spark native scheduler does something similar to that, you can probably start by looking at the source code.
If you don't want to use Kubernetes you can either use Mesos or YARN that implement DRF natively.