Course contentsShow
← Cloud and Infrastructure
Lesson 1325 of 2,98926. Kubernetes StorageFree lesson

Raw Block Volumes

Provisioning and consuming raw block devices instead of filesystems using volumeMode: Block in PVCs.

Raw Block Volumes

What You'll Learn

You'll learn how to provision and consume raw block devices in Kubernetes instead of traditional filesystems, giving applications direct access to storage blocks.

Understanding Raw Block Volumes

Until now, you've worked with volumes that present a filesystem to your pods—think of directories and files you can navigate. But some applications (like databases or storage systems) prefer working with raw block storage: direct access to storage blocks without the overhead of a filesystem layer.

Imagine the difference between receiving a pre-organized filing cabinet (filesystem) versus getting raw, blank filing drawers you can organize exactly how you want (block device). The second option gives specialized applications more control and often better performance.

Provisioning Raw Block Volumes

To request a raw block volume instead of a filesystem, you set volumeMode: Block in your PersistentVolumeClaim:

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: raw-block-pvc
spec:
  accessModes:
    - ReadWriteOnce
  volumeMode: Block  # Request raw block device
  resources:
    requests:
      storage: 10Gi
  storageClassName: fast-ssd

Consuming in Pods

Instead of volumeMounts, you use volumeDevices to attach the block device:

apiVersion: v1
kind: Pod
metadata:
  name: block-consumer
spec:
  containers:
  - name: app
    image: myapp
    volumeDevices:  # Not volumeMounts!
    - name: data
      devicePath: /dev/xvda  # Appears as a block device
  volumes:
  - name: data
    persistentVolumeClaim:
      claimName: raw-block-pvc

The application now sees /dev/xvda as a raw block device it can read/write directly.

Key Takeaway: Raw block volumes using volumeMode: Block provide direct block device access to applications that need low-level storage control, bypassing filesystem overhead for better performance in specialized workloads.