8 Key Facts About Kubernetes Volume Group Snapshots Now GA in v1.36

Kubernetes v1.36 is here, and one of its standout features is the General Availability (GA) of volume group snapshots. This capability, which started as an alpha in v1.27, has evolved through beta stages to become a production-ready tool for taking crash-consistent snapshots across multiple volumes. Whether you're managing stateful applications or planning disaster recovery, volume group snapshots promise to simplify bulk backup operations while ensuring data integrity.

In this article, we break down the eight essential things you need to know about this GA release, from the underlying APIs to practical use cases. Let's dive in.

1. What Are Volume Group Snapshots, Exactly?

Volume group snapshots are point-in-time copies of multiple persistent volumes taken simultaneously. Unlike individual volume snapshots—which capture a single volume at a moment in time—group snapshots ensure write order consistency across all volumes in the set. This is crucial for applications that span multiple volumes, such as a database where data resides in one volume and transaction logs in another. If you snapshot them separately, you risk restoring an inconsistent state. Group snapshots eliminate that risk by creating a unified, crash-consistent recovery point.

Under the hood, Kubernetes uses a label selector to identify the PersistentVolumeClaims (PVCs) belonging to the group. The storage system then captures a snapshot for each volume at exactly the same instant. These snapshots can later be used to restore volumes to new PVCs or to rehydrate existing volumes—much like a traditional backup but with guaranteed consistency.

2. How Does It Differ from Individual Volume Snapshots?

Kubernetes already had the VolumeSnapshot API for individual volumes, which works fine for single-volume applications. But for multi-volume workloads, the old approach either required manual orchestration—quiescing the application, snapshotting each volume one by one—or risked inconsistencies. Volume group snapshots solve this by providing a single API call that coordinates the snapshot across all selected PVCs. The result is a set of snapshots that can be restored together, preserving the application's state as it was at the snapshot time.

Importantly, you don't need to pause or quiesce the application anymore in many cases. The storage system handles the crash consistency internally, making the process faster and less disruptive. This is a significant improvement for high-availability environments where downtime must be minimized.

3. The Core API Types You Need to Know

Volume group snapshots are managed through three custom resource definitions (CRDs):

• VolumeGroupSnapshot — Created by a user (or automation) to request a group snapshot. You define the PVCs via label selectors and optional snapshot class.
• VolumeGroupSnapshotContent — Automatically generated by the snapshot controller when a VolumeGroupSnapshot is dynamically provisioned. It represents the actual snapshot data on the storage backend and is bound to the VolumeGroupSnapshot.
• VolumeGroupSnapshotClass — Defines the driver-specific parameters for creating group snapshots. This is analogous to the VolumeSnapshotClass used for individual snapshots.

Together, these APIs form the building blocks for creating, managing, and restoring group snapshots. They follow the same patterns as existing Kubernetes snapshot APIs, so if you're already familiar with VolumeSnapshot, you'll find the group version intuitive.

4. It's a CSI-Only Feature

Volume group snapshots are only available for CSI (Container Storage Interface) drivers. In-tree volume plugins and older drivers (like those based on FlexVolume) are not supported. This is because the group snapshot functionality relies on CSI's ability to orchestrate snapshots across multiple volumes in a single operation. If you're using a CSI driver that supports group snapshots (check your vendor's compatibility), you can enable the feature right away. For others, adoption may require migrating to a CSI-compliant driver.

The GA release ensures that CSI drivers implementing the GroupSnapshot capability can be fully utilized without workarounds. The Kubernetes community has worked closely with storage vendors to finalize the CSI spec extensions, so support should be broad and reliable.

5. Crash Consistency: The Killer Feature

The primary benefit of volume group snapshots is crash consistency. When snapshots are taken at different times across volumes, an application might see a state where data is written but logs are not, or vice versa. This can lead to data corruption upon restore. A crash-consistent snapshot captures all volumes at the exact same instant, mirroring the state that would exist if the system suddenly crashed.

This is especially valuable for databases like PostgreSQL or MySQL that maintain data files, WAL logs, and config files on separate volumes. With group snapshots, you can restore an entire database system to a consistent point, bypassing the need for complex recovery procedures. It also simplifies backup automation—no more scripting to orchestrate individual snapshots with application quiescence.

6. How to Create a Volume Group Snapshot in Practice

Creating a group snapshot is straightforward. First, ensure your CSI driver supports it and that you've installed the required CRDs (they're part of the Kubernetes snapshot controller). Then, define a VolumeGroupSnapshot manifest:

apiVersion: groupsnapshots.storage.k8s.io/v1alpha1
kind: VolumeGroupSnapshot
metadata:
  name: my-group-snapshot
spec:
  volumeGroupSnapshotClassName: csi-group-snapshot-class
  source:
    selector:
      matchLabels:
        app: my-database

This example targets all PVCs labeled app: my-database. The CSI driver will snapshot each volume in the group simultaneously. After creation, you can check the status via kubectl describe—the resulting VolumeGroupSnapshotContent objects will be created automatically. You can then use them to restore volumes later by referencing the group snapshot in a new PVC definition.

7. Restoring from a Group Snapshot

Restoring volumes from a group snapshot is similar to restoring individual snapshots, but with a key difference: you can restore all volumes from the group at once. To do so, you specify the VolumeGroupSnapshot as the data source in a new PVC. Kubernetes will provision new volumes pre-populated with the snapshot data. This is often done in a secondary namespace or cluster for testing or disaster recovery.

Alternatively, you can restore individual volumes from the group by referencing the corresponding VolumeGroupSnapshotContent directly. This flexibility is useful when you only need a subset of the data. Keep in mind that the crash consistency guarantee applies only when you restore the entire group together. Mixing restored volumes with live ones may break consistency.

8. When to Use Volume Group Snapshots

Volume group snapshots shine in several scenarios:

• Multi-volume applications like databases, CMS platforms, or analytics tools that store data across multiple PVCs.
• Disaster recovery where you need to capture a consistent state of the entire application at regular intervals.
• CI/CD pipelines that snapshot test environments for reproducibility.
• Migration between clusters or storage backends—snapshots can be restored into new volumes.

However, they are not a replacement for application-aware backups. For workloads that require transactional consistency beyond crash consistency (e.g., quiescing the database engine), you may still need to combine group snapshots with application-level hooks. Also, not all CSI drivers support group snapshots; verify compatibility before designing your backup strategy.

With the GA release in Kubernetes v1.36, volume group snapshots are now ready for production workloads. They represent a significant step forward in storage management, making it easier to protect complex stateful applications without sacrificing data integrity.