Containers · in-mumbai-1 · in-noida-1

Kubernetes with a ₹0 control plane

EKS charges ₹6,935 a month per cluster before you schedule a single pod — ₹83,220 a year, per cluster. Our conformant control plane is free, HA optional, upgrades are one click, and you pay only for the workers.

Create a cluster — ₹25,000 credit View capabilities ↓

Control plane ₹0Workers from ₹999/moUpgrades one-clickAutoscale min/max nodesCNI + CSI integrated

Capabilities

Managed where it should be, standard where it must be

Cluster in minutes

Pick a Kubernetes version, worker size and count — the control plane, networking and node pool come up ready for kubectl.

Node autoscaling

Set min/max workers and the autoscaler adds nodes when pods go pending, then trims them when the rush passes. Pair with HPA inside the cluster.

One-click upgrades

Move to a new Kubernetes version with a rolling upgrade — masters first, then cordon-and-drain across workers, respecting your PodDisruptionBudgets.

Storage that follows pods

PersistentVolumeClaims provision NVMe block volumes through the CSI driver — expandable online, snapshottable on schedule.

Real LoadBalancer Services

type=LoadBalancer provisions our flat-rate balancer automatically — a public VIP without annotations gymnastics or LCU anxiety.

Private clusters in your VPC

Deploy into a VPC tier, restrict the API endpoint with ACLs, and keep pod-to-database traffic entirely on private Indian networks.

Under the hood

Upstream Kubernetes, CloudStack-native integrations

Clusters run unmodified upstream, CNCF-conformant Kubernetes builds — currently the latest supported minors (e.g. 1.32 / 1.33), with new versions added shortly after upstream release. No fork, no vendored kubectl.

The cloud-controller-manager wires Services to our load balancers; the CSI driver turns PVCs into NVMe volumes. Your manifests and Helm charts don't know they left AWS.

kubeconfig download with scoped credentials — kubectl works in 30 seconds
containerd runtime, standard CNI networking between pods
HA control plane option across separate hosts
Cluster state, images and PVs all resident in India
Terraform and API parity for cluster lifecycle

HelmIngress-NGINXcert-managerArgoCDPrometheusGrafanacontainerdkubectl

Frequently asked questions

Is this real, conformant Kubernetes?+

Yes — unmodified upstream builds passing CNCF conformance. kubectl, Helm, operators and your existing manifests work as-is; the only cloud-specific pieces are the storage and load-balancer integrations, which follow standard CSI and cloud-provider interfaces.

How can the control plane be free?+

Master nodes are small and we absorb their cost as part of the managed service — the economics work because you run workers with us. There's no per-cluster meter ticking the way EKS's $0.10/hour does.

How do upgrades work in practice?+

One click starts a rolling upgrade: control plane first, then workers cordon-and-drain one at a time, honouring PodDisruptionBudgets. Test on a staging cluster first — that's free too, since only workers cost money.

Can the cluster be fully private?+

Yes. Deploy into a VPC tier, lock the API endpoint down with network ACLs, and expose only what you choose through a load balancer. East-west traffic never touches the public internet.

What about GPU nodes for ML workloads?+

GPU worker pools are available through our GPU cloud line — mixed CPU/GPU node pools in one cluster. Contact us with your card and driver requirements and we'll size it.

Pairs well with

Complete the stack

Cloud ServersThe D-series machines your workers run on→Block StoragePersistent Volumes at ₹5/GB via CSI→Load BalancerFlat-rate ingress for your Services→