Deploying a Multi-Node ScyllaDB Cluster on Kubernetes

itflow

🌀 Deploying a Multi-Node ScyllaDB Cluster on Kubernetes

ScyllaDB offers native support for Kubernetes via its Scylla Operator. This guide will walk you through deploying a multi-node ScyllaDB cluster on Kubernetes using the Scylla Operator.

🚀 Prerequisites

Kubernetes cluster (Minikube, kind, or cloud-managed like GKE, EKS, AKS)
kubectl configured and connected to your cluster
Helm (v3+)
Optional: k9s for terminal-based Kubernetes management

⚙️ Step-by-Step Guide

1. Create a Namespace

kubectl create namespace scylla

2. Install Scylla Operator via Helm

helm repo add scylla-operator https://scylla-operator-charts.storage.googleapis.com/stable
helm repo update

helm install scylla-operator scylla-operator/scylla-operator \
  --namespace scylla

Verify the operator is running:

kubectl get pods -n scylla

3. Deploy a Scylla Cluster

Create a file called scylla-cluster.yaml:

apiVersion: scylla.scylladb.com/v1
kind: ScyllaCluster
metadata:
  name: scylla-cluster
  namespace: scylla
spec:
  version: 5.4.0
  datacenter:
    name: dc1
    racks:
      - name: rack1
        members: 3
        resources:
          requests:
            cpu: 1
            memory: 2Gi

⚠️ Make sure your cluster has enough resources to schedule 3 pods with the above CPU/Memory requests.

Apply the config:

kubectl apply -f scylla-cluster.yaml

4. Monitor the Cluster

Check the pods:

kubectl get pods -n scylla -l scylla.scylladb.com/cluster=scylla-cluster

Check the status of the cluster:

kubectl describe ScyllaCluster scylla-cluster -n scylla

🧪 Accessing ScyllaDB

You can port-forward to access CQL:

kubectl port-forward svc/scylla-cluster-dc1-service -n scylla 9042:9042

Then connect using cqlsh:

cqlsh localhost 9042

Make sure you have Cassandra tools installed locally to use cqlsh.

🔄 Scaling the Cluster

To scale up to 5 nodes, update your scylla-cluster.yaml:

        members: 5

Apply the updated manifest:

kubectl apply -f scylla-cluster.yaml

🧹 Cleanup

To delete everything:

kubectl delete ScyllaCluster scylla-cluster -n scylla
helm uninstall scylla-operator -n scylla
kubectl delete namespace scylla

📚 Useful Resources

✅ Conclusion

You now have a multi-node ScyllaDB cluster running in Kubernetes, scalable and production-ready. With the Scylla Operator, managing clusters becomes declarative and Kubernetes-native.

For advanced use, consider:

Setting up Monitoring Stack with Prometheus + Grafana
Adding NodeSelectors, Tolerations, or Persistent Volumes
Running across multiple datacenters or regions

Happy scaling! 🏗️📈