AWS, Azure & GCP · Cloud across providers
What do managed Kubernetes services handle for you?
Short answer
Managed Kubernetes runs the control plane and handles its availability and upgrades, while you own node pools, workloads, networking, RBAC and add-ons. It reduces control-plane operations but not cluster operations.
EKS, AKS and GKE run the API server, etcd and scheduler for you, usually across zones with a service-level commitment. That removes the most fragile part of self-managed Kubernetes, but the cluster is still a platform you operate.
What you still own
- Node pools, instance types, autoscaling and OS image patching
- Workload manifests, resource requests and limits, and rollout strategy
- Cluster networking, ingress, DNS and network policy
- RBAC, workload identity, secret delivery and admission policy
- Add-ons such as ingress controllers, certificate management and observability
apiVersion: apps/v1
kind: Deployment
metadata:
name: api
spec:
replicas: 3
selector:
matchLabels:
app: api
template:
metadata:
labels:
app: api
spec:
containers:
- name: api
image: registry.example.com/api:1.4.2
resources:
requests:
cpu: 250m
memory: 256Mi
limits:
memory: 512MiThe control plane being managed does not make workloads highly available. You still need multiple replicas, pod disruption budgets, topology spread, readiness probes and a rollout strategy. Cluster upgrades also need planning because node and API versions drift.
Follow-ups you should expect
Interviewers use these to check whether the first answer was understanding or memorisation.
- 1What is the difference between control-plane and node responsibilities?
- 2How do you upgrade a cluster without downtime?
- 3Why do resource requests and limits matter for scheduling?
- 4When would you choose a simpler container service over Kubernetes?
Say it in the interview
A managed Kubernetes service runs the control plane and keeps it available, but I still own node pools, workloads, networking, RBAC and add-ons. I use multiple replicas, readiness probes and disruption budgets so the cluster being managed does not imply the workload is resilient. I also plan upgrades and watch the total cost, because control-plane fees are only part of the bill.
Use this as a model, not a script. Saying it in your own words is what makes it survive follow-up pressure.
Related questions
Reading is step one
Now say it out loud.
Understanding an answer and delivering it under pressure are different skills. Rehearse this topic in a live AWS, Azure & GCP mock interview and get scored feedback on depth, clarity and confidence.