Velero 1.18: Concurrent Backups, Cache Volumes, and More

Shubham Pampattiwar
March 06, 2026

We are pleased to announce the release of Velero v1.18. This release brings significant improvements in concurrency, performance, and observability, with contributions from engineers across multiple organizations.

Full list of changes can be found here

Release Highlights

Concurrent Backup Processing

Velero can now process multiple backups concurrently. This is a major usability improvement for multi-tenant environments – backups submitted by different users or teams run simultaneously without interfering with each other.

Previously, backups were serialized, meaning a long-running backup would block all other pending backups. With concurrent processing, backup throughput scales with available resources.

See the design document for details.

Cache Volume Support for Data Movers

Velero 1.18 allows users to configure cache volumes for data mover pods during restore operations for both CSI snapshot data movement and fs-backup. This solves several real-world problems:

  • Data mover pods failing when a pod’s ephemeral disk is limited
  • Multiple data mover pods failing to run concurrently on a single node due to disk constraints
  • Combined with backup repository cache limit configuration, appropriately sized cache volumes improve restore throughput

See the design document for details.

Incremental Backup Size Reporting

Users can now observe the incremental size of data mover backups for CSI snapshot data movement and fs-backup. This provides visibility into data reduction from incremental backups, helping teams understand and optimize their backup storage usage.

Wildcard Namespace Filtering

Velero now supports Glob regular expressions for namespace filters during backup and restore. This allows users to filter namespaces in batch – for example, backing up all namespaces matching team-* or excluding test-* namespaces.

VolumePolicy Enhancements

VolumePolicy receives two improvements in this release:

  • PVC Phase support: Users can now filter volumes by PVC phase, enabling actions like skipping PVCs in Pending or Lost status from backups to avoid failures caused by unbound volumes.
  • VolumeGroupSnapshot integration: Volume policies now apply to VolumeGroupSnapshot PVC filtering, building on the VolumeGroupSnapshot support introduced in v1.17.

Scalability and Resiliency

Prevent Velero Server OOM for Large Backup Repositories

Some backup repository operations are now executed outside the Velero server process, preventing OOM kills when working with large repositories.

VolumePolicy Performance

VolumePolicy evaluation has been optimized for environments with large numbers of pods and PVCs, resulting in significantly improved performance through a PVC-to-Pod cache that avoids redundant lookups.

Events for Data Mover Pod Diagnostics

Events are now recorded in data mover pod diagnostics, giving users more information for troubleshooting when data mover pods fail.

Breaking Changes

Deprecation of PVC Selected Node Feature

Per the Velero deprecation policy, the PVC selected node feature is deprecated in v1.18. Velero now handles PVC selected-node annotations automatically, so no user action is required.

Community Contributions

This release includes contributions from across the Velero community. Thank you to everyone who contributed:

Join the Community

Related Content
Velero 1.17: Volume Group Snapshots, Modernized fs-backup, and Windows Support
Velero Joins the CNCF Sandbox
Velero Joins the CNCF Sandbox
Velero has been accepted into the Cloud Native Computing Foundation as a Sandbox project, bringing Kubernetes-native backup and disaster recovery under vendor-neutral, community-driven governance.
Getting Started

To help you get started, see the documentation.

Documentation