Feature Overview (aka. Goal Summary)  

Support the ability to limit network bandwidth of a cluster namespace.
 

Goals (aka. expected user outcomes)

Provide an admin-controlled bandwidth limit to a set of pods within a namespace. Without this capability, each pod can otherwise arbitrarily set their own bandwidth limits, and potentially consume all available bandwidth to the cluster to the detriment of other pods.
 

Requirements (aka. Acceptance Criteria):

The cluster admin can specify a new bandwidth limit parameter (e.g. networkBandwidth: 100M) for the namespace in a fashion similar to:

apiVersion: v1
kind: ResourceQuota
metadata:
name: quota
spec:
hard:
cpu: "1000"
memory: 200Gi
pods: "10"
networkBandwidth: 100M

With the admin-configured ResourceQuota similar to the above, individual users can split the allocated bandwidth amongst their pods as follows:

PodA: 50M
PodB: 30M
PodC: 20M

Use Cases (Optional):

As an OpenShift cluster administrator, I require the ability to control the total bandwidth consumption of related (same namespace) pod owners.
 

Questions to Answer (Optional):

Out of Scope

Limits based upon CPU and/or memory.

Background

Customer Considerations

This feature is an adoption blocker for some customers.

Documentation Considerations

Interoperability Considerations