Introduction to Persistent Volume Management in Kubernetes
By design, Kubernetes Pods are temporary. If a Pod crashes, gets rescheduled to another server, or scales down, everything saved inside its local container memory is instantly destroyed. While this is great for stateless applications like web servers, it is a disaster for databases, user uploads, or any application where your data shouldn't be lost.
To solve this, Kubernetes uses a storage architecture that separates the physical storage from the Pods that use it. In this tutorial, we will learn how to configure Persistent Volumes (PVs), Persistent Volume Claims (PVCs), and dynamic StorageClasses so your application data survives pod restarts safely.
- The difference between Persistent Volumes (PVs) and Persistent Volume Claims (PVCs).
- How a StorageClass automates the creation of physical disks dynamically.
- How to write YAML to request storage and attach it to a Pod.
- The lifecycle and reclaim policies of Kubernetes storage.
1. The Core Concepts of Kubernetes Storage
Think of Kubernetes storage like renting an apartment. The Persistent Volume (PV) is the actual physical apartment building. The Persistent Volume Claim (PVC) is your rental application saying "I need a 2-bedroom apartment." The StorageClass is the property manager who automatically builds new apartments when people request them.
By keeping the "request" (PVC) separate from the "actual disk" (PV), developers don't need to know if the underlying storage is an AWS EBS volume, an Azure Disk, or a local hard drive on a bare-metal server. They just ask for storage, and Kubernetes handles the rest.
| Concept | What it represents | Who manages it? |
|---|---|---|
| StorageClass (SC) | The blueprint or "property manager". It defines what kind of storage to provision (e.g., fast SSD or cheap HDD) and automates the creation of PVs. | Cluster Administrator |
| Persistent Volume (PV) | The actual piece of storage in the cluster (e.g., a 10GB block of a hard drive). It exists independently of any Pod. | Admin or StorageClass (Auto) |
| Persistent Volume Claim (PVC) | A request for storage by a developer. E.g., "I need 5GB of ReadWrite storage." Kubernetes matches this claim to an available PV. | Application Developer |
2. Practical Implementation & Code
Let's look at how to request storage dynamically. We will create a PVC, and then we will create an NGINX Pod that mounts this storage. Because we are using a dynamic StorageClass (which is standard on local setups like Minikube, Docker Desktop, or cloud providers), we do not need to manually create the PV—the StorageClass will do it for us.
Creating the Storage Request and the Pod
Example 1: pvc-and-pod.yaml
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
name: my-app-pvc
spec:
accessModes:
- ReadWriteOnce # Only one node can mount this volume as read-write
resources:
requests:
storage: 2Gi # We are asking for 2 Gigabytes of storage
---
apiVersion: v1
kind: Pod
metadata:
name: my-storage-pod
spec:
containers:
- name: nginx-container
image: nginx:latest
volumeMounts:
- mountPath: "/usr/share/nginx/html/data" # Where the storage appears inside the container
name: my-persistent-storage
volumes:
- name: my-persistent-storage
persistentVolumeClaim:
claimName: my-app-pvc # This links the Pod to the PVC created above
Expected output (when running 'kubectl apply -f pvc-and-pod.yaml')
persistentvolumeclaim/my-app-pvc created pod/my-storage-pod created
claimName inside the Pod's volume definition MUST exactly match the name of the PersistentVolumeClaim. If there is a typo, the Pod will remain in a "Pending" state forever.3. Architecture / Workflow Visualized
Let's visualize how the request flows from the Pod down to the actual physical hard drive. Notice how the PVC acts as the middleman protecting the Pod from needing to know the hardware details.
Kubernetes Dynamic Storage Provisioning Workflow
When the developer submits the PVC, the StorageClass detects the request and automatically provisions the PV, completing the binding process.
4. Step-by-Step Execution Breakdown
When you apply the YAML file above, Kubernetes performs a specific sequence of actions behind the scenes to secure your data.
/usr/share/nginx/html/data).5. Hands-On Practice
Ready to test your storage knowledge? Fire up your local cluster (Minikube or Docker Desktop) and try these exercises.
Write a YAML file to create a PVC named `database-pvc` requesting 5Gi of storage with the `ReadWriteOnce` access mode. Apply it to your cluster and check its status.
Hint: Use kubectl get pvc. The status should say Bound if your local cluster has dynamic provisioning enabled.
Deploy a busybox Pod that mounts the PVC you just created to `/data`. Run a command inside the pod to write a text file to that folder.
Expected result: You can run kubectl exec -it [pod-name] -- sh and type echo "Hello Data" > /data/test.txt.
Delete the Pod you created in Exercise 2 (do NOT delete the PVC). Create a brand new Pod that mounts the same PVC to `/new-data`. Execute into the new Pod and check if `test.txt` survived.
Hint: Because the PVC was never deleted, the new Pod will attach to the existing physical storage. Your file will be waiting for you!
6. Interview Questions
Storage is a critical topic for Kubernetes Administrator interviews. Here is what hiring managers want to know.
Answer: In static provisioning, a cluster administrator must manually create PVs in advance. Developers then create PVCs to claim them. In dynamic provisioning, the administrator sets up a StorageClass. When a developer submits a PVC, the StorageClass automatically creates the PV and physical storage on-demand, saving manual effort.
Answer: There are three main modes. ReadWriteOnce (RWO): The volume can be mounted as read-write by a single node. ReadOnlyMany (ROX): The volume can be mounted read-only by many nodes simultaneously. ReadWriteMany (RWX): The volume can be mounted as read-write by many nodes simultaneously (usually requires network-attached storage like NFS).
Conclusion
Mastering Persistent Volumes (PVs) and Persistent Volume Claims (PVCs) is essential for running serious, stateful applications in Kubernetes. By separating the storage request from the storage implementation, Kubernetes allows developers to build portable applications that work identically on a local laptop, on AWS, or on on-premises servers. Remember to always double-check your Reclaim Policies before deleting PVCs in production!
Official Documentation
Upgradix by Shrash Studio – Learn Technology. Build Your Future.