Kubernetes Persistence: Taming Multiple Claim Types in Your Cluster

8 August 2024

Understanding Persistent Volumes and Claims in Kubernetes

In the world of Kubernetes, managing storage is a crucial aspect of deploying applications. The Persistent Volume (PV) and Persistent Volume Claim (PVC) mechanisms were introduced to provide a way to manage this storage need. A PV represents a resource on the cluster that can be used to store data, while a PVC is a request for that resource. This separation allows for greater flexibility in managing resources across different applications.

The Challenge of Multiple Claim Types

While the basic model provides significant benefits, it often fails to meet the needs of complex storage scenarios. One common requirement is the ability to use multiple claim types within a single application or even across different parts of the same cluster. This can involve combining claims for different sizes (e.g., small and large volumes), types (e.g., SSDs and HDDs), and purposes (e.g., data storage, logging).

Solution: Using Multiple Persistent Volume Claim Types

To tackle this challenge in Kubernetes, you can leverage the ability to specify multiple claim types within a single PVC request. This involves defining your PVC with a specification that includes multiple volumes of different sizes or types.

Example YAML for Multiple Claim Types

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: multi-claim-example
spec:
  accessModes:
    - ReadWriteOnce
  resources:
    requests:
      storage: 10Gi
  volumeSelector:
    matchLabels:
      type: "small"
      size: "medium"
  extra:
    - size: 20Gi
      type: "large"

In this example, the multi-claim-example PVC is specified to request a total of 10Gi storage. However, it also includes an additional request for storage that matches labels indicating a “large” volume with a specific size of 20Gi. This setup allows you to handle multiple claim types within a single PVC request.

Deployment and Configuration

To deploy this configuration in your Kubernetes cluster, ensure that the Persistent Volume (PV) objects are created to match the requirements specified in your PVC. These PVs should be tagged or labeled with the correct type and size to correspond with the PVC’s volume selector.

Example YAML for Persistent Volume

apiVersion: v1
kind: PersistentVolume
metadata:
  name: pv-small-medium
spec:
  capacity:
    storage: 10Gi
  accessModes:
    - ReadWriteOnce
  type: local
  nodeAffinity:
    required:
      nodeSelectorTerms:
        - matchExpressions:
            - key: kubernetes.io/hostname
              operator: In
              values:
                - <node-name>

In this example, a PV named pv-small-medium is created with a capacity of 10Gi, matching the “small” and “medium” volume type labels specified in your PVC. Ensure that you replace <node-name> with the actual name of the node where this PV will be hosted.

Conclusion

Utilizing multiple Persistent Volume claim types within Kubernetes offers greater flexibility for managing storage across applications or parts of a cluster. By specifying different sizes, types, and purposes of claims within a single PVC request, you can optimize your resource usage more effectively. The examples provided in YAML format demonstrate how to configure both the PVC and PV objects to meet these requirements in your cluster.

Poespas Blog