10 Cloud Container Backup Tools You Must Know
10 cloud container backup tools: AWS Backup for EKS, Azure Backup for AKS, Backup for GKE, OADP, Velero, Kasten, CloudCasa, Trilio, Portworx, registries.
Cloud container backup tools protect three things: cluster state (Kubernetes objects), persistent volume data, and the container images your workloads pull. The hyperscalers now have managed options (AWS Backup for EKS, Azure Backup for AKS, Backup for GKE), while Velero, Kasten, CloudCasa, Trilio, Portworx Backup and OADP cover portability. Registry replication rounds out the set.
This post is organised by cloud and by what each tool protects, so you can pick the native option for the platform you run on and know when to add a portable tool. For a deeper enterprise feature comparison see our 12 Kubernetes backup tools for enterprise data, and for failover planning read Kubernetes disaster recovery tools.
Key takeaways
- A container backup needs cluster resources, volume data and images. Many teams back up only one of the three.
- AWS Backup added native Amazon EKS support in November 2025, joining Azure Backup for AKS and Backup for GKE as first-party managed services.
- Native services are simplest inside one cloud. Portable tools such as Velero and Kasten matter for multi-cloud, hybrid or migration use.
- Images are often forgotten. Cross-region registry replication or a second registry protects against losing the images needed to restore.
- A backup you have never restored is untested. Schedule restore drills into a separate cluster.
What a container backup must include
| Layer | Examples | Why it matters |
|---|---|---|
| Cluster state | Deployments, Services, ConfigMaps, Secrets, CRDs, RBAC | Without it you rebuild the application by hand, even if the data survived |
| Persistent data | PersistentVolumes on EBS, Azure Disk, Persistent Disk, EFS, Azure Files | Databases and file stores live here |
| Images | Repositories in ECR, ACR, Artifact Registry, Harbor | A restored Deployment fails with ImagePullBackOff if the image is gone |
| Outside config | Cloud IAM roles, load balancers, DNS, secrets managers | Often outside the cluster and outside the backup tool |
GitOps helps with the first layer: if your manifests live in Git and Argo CD or Flux applies them, cluster state can be recreated from the repository. Backups are still needed for data and for anything created outside Git.
Managed backup services from the cloud providers
1. AWS Backup for Amazon EKS
AWS announced native EKS support in AWS Backup in November 2025. It backs up EKS cluster state together with persistent data from supported AWS storage, managed through the same backup plans, vaults and policies used for EC2, RDS and EBS. AWS added restore into a new cluster built from the backed-up configuration, later added support for logically air-gapped vaults (March 2026) and reported faster cluster-state backups in May 2026.
Best for: teams already using AWS Backup and AWS Organizations backup policies who want EKS in the same governance model. Check the AWS Backup feature availability page for supported storage types and Regions.
2. Azure Backup for AKS
Azure Backup protects AKS clusters through a backup extension installed in the cluster and a Backup vault. It captures cluster resources and persistent volumes backed by Azure Disk, with policies for schedule and retention and options to restore to the original or another cluster. Microsoft documents its supported scenarios and limits on Microsoft Learn.
Best for: AKS estates that want backup managed in the Azure portal alongside VM and database backups, with Azure Policy and Backup Center reporting.
3. Backup for GKE
Google Cloud's Backup for GKE is a managed service that backs up and restores GKE workloads, including Kubernetes resources and Persistent Disk volumes. You define backup plans per cluster and restore plans that control which namespaces or applications are restored and how conflicts are handled.
Best for: GKE users who want backup scheduling, retention and restores handled by Google Cloud with IAM and audit logging.
4. OpenShift API for Data Protection (OADP)
Red Hat OpenShift ships OADP, an operator based on the upstream Velero project. It backs up and restores OpenShift namespaces, resources and volumes to object storage, and works on self-managed OpenShift as well as managed versions such as ROSA and ARO.
Best for: OpenShift users who want a Red Hat supported path instead of installing Velero themselves.
Portable backup tools that work across clouds
5. Velero
Velero is the open-source standard for Kubernetes backup. It stores resource backups in object storage and protects volumes either through cloud snapshots via plugins or through file-system backup with Kopia. It is free to run, and many commercial products build on it.
# Back up one namespace, then restore it into a test cluster
velero backup create shop-daily --include-namespaces shop --ttl 720h
velero backup describe shop-daily --details
velero restore create --from-backup shop-daily
Best for: teams comfortable operating their own tooling, and anyone migrating workloads between clusters or clouds.
6. Veeam Kasten
Kasten, part of Veeam since its 2020 acquisition, is a commercial Kubernetes data protection platform with an application-centric UI, policy-based backups, application-consistent hooks for databases, and ransomware-focused features such as immutable backups. It runs inside the cluster and supports the major managed Kubernetes services.
Best for: enterprises that want a supported product with dashboards and compliance reporting.
7. CloudCasa
CloudCasa offers Kubernetes backup and recovery as a SaaS service with a self-hosted option. It works with EKS, AKS and GKE and can manage existing Velero installations, which helps teams that started with Velero and now want central management.
Best for: organisations with many clusters across clouds that want a hosted control plane.
8. Trilio for Kubernetes
Trilio provides application-centric backup and recovery for Kubernetes and OpenShift, with an operator-based install and restore across clusters. It is commonly evaluated by OpenShift and telecom users.
Best for: OpenShift-heavy estates that want an alternative to OADP with commercial support.
9. Portworx Backup
Portworx Backup, from Pure Storage, protects Kubernetes applications and data across clusters and clouds. It pairs naturally with Portworx Enterprise storage but can also protect volumes on other CSI drivers.
Best for: teams already using Portworx for container storage.
Do not forget the images
10. Registry replication (ECR, ACR, Artifact Registry, Harbor)
Restoring workloads only works if their images can still be pulled. Each major registry offers a way to keep copies elsewhere:
- Amazon ECR supports cross-Region and cross-account replication rules.
- Azure Container Registry supports geo-replication on the Premium tier.
- Google Artifact Registry offers multi-region repositories and remote repositories.
- Harbor, a CNCF graduated project, has policy-based replication to other registries.
Pin images by digest in production manifests so a restore pulls exactly the same build. See container registry practices for security and compliance for related controls.
Comparison table
| Tool | Type | Clouds | Protects | Cost model |
|---|---|---|---|---|
| AWS Backup for EKS | Managed service | AWS | Cluster state, supported volumes | Pay per use (AWS pricing) |
| Azure Backup for AKS | Managed service | Azure | Cluster resources, Azure Disk volumes | Pay per use (Azure pricing) |
| Backup for GKE | Managed service | Google Cloud | Resources, Persistent Disk volumes | Pay per use (Google Cloud pricing) |
| OADP | Operator (Velero based) | Any OpenShift | Resources, volumes | Included with OpenShift subscription |
| Velero | Open source | Any | Resources, volumes | Free; you run it |
| Veeam Kasten | Commercial | Any | Applications, volumes | Licensed (free tier available per vendor) |
| CloudCasa | SaaS or self-hosted | Any | Resources, volumes | Subscription |
| Trilio | Commercial | Any, strong on OpenShift | Applications, volumes | Licensed |
| Portworx Backup | Commercial | Any | Applications, volumes | Licensed |
| Registry replication | Registry feature | Per registry | Container images | Storage and transfer charges |
Check each vendor's current pricing page; we do not quote prices because they change often.
How to choose
- One cloud, managed Kubernetes: start with that cloud's native service. Governance, IAM and billing are already in place.
- Multi-cloud or hybrid: add a portable tool (Velero, Kasten, CloudCasa) so backups from one platform can restore on another.
- OpenShift: OADP first; Trilio or Kasten if you need more features or support.
- Databases on Kubernetes: make sure the tool supports pre- and post-backup hooks or operator-native backups for application consistency.
- Ransomware concerns: require immutable or air-gapped backup storage, such as S3 Object Lock or a logically air-gapped vault, and separate credentials.
A minimal restore drill
- Pick one namespace with a stateful app.
- Restore last night's backup into a separate test cluster or namespace.
- Confirm pods start, images pull and the app reads its data.
- Record how long it took. That number is your real recovery time, not the one in the design document.
- Repeat quarterly and after major upgrades.
Our Kubernetes backup strategies article covers scheduling, retention and the 3-2-1 rule in more depth, and Kubernetes storage options explains the volume types these tools protect. Backup and restore of cluster state, including etcd snapshots, is part of the CKA syllabus covered in our CKA training in Pune.
Application-consistent backups for stateful workloads
A volume snapshot taken while a database is writing can capture a half-finished transaction. The restore may then need crash recovery, or in the worst case lose data. There are three common ways to get a consistent copy:
- Backup hooks. Velero, Kasten and similar tools can run a command inside a pod before and after the snapshot, for example to flush or freeze writes and then release them.
- Operator-native backups. Database operators such as CloudNativePG for PostgreSQL can stream backups to object storage with point-in-time recovery. Use these for the database and the cluster backup tool for everything else.
- Managed databases. Running the database outside the cluster, on RDS, Azure Database or Cloud SQL, moves backup responsibility to the provider's backup features.
A pre-backup hook in Velero is set with pod annotations:
metadata:
annotations:
pre.hook.backup.velero.io/container: postgres
pre.hook.backup.velero.io/command: '["/bin/bash", "-c", "psql -U postgres -c CHECKPOINT"]'
pre.hook.backup.velero.io/timeout: 60s
Security controls for backup data
Backups hold Secrets, customer data and the blueprint of your systems, which makes them a target. Apply the same care as production:
| Control | How to apply it |
|---|---|
| Separate account or subscription for backup storage | A compromise of the workload account should not allow deletion of backups |
| Immutability | S3 Object Lock, Azure immutable vaults, Cloud Storage retention policies or a logically air-gapped vault |
| Encryption | Encrypt at rest, with customer-managed keys where policy requires, and restrict key access |
| Least privilege for the backup tool | Grant only the cluster and storage permissions the tool needs |
| Monitoring | Alert on failed backups, missed schedules and unexpected deletions |
The reasoning behind immutable copies is explained in how immutable backups support ransomware-resistant architectures.
Retention and recovery targets
Set two numbers for each application before choosing schedules: the recovery point objective (how much data you can afford to lose) and the recovery time objective (how long you can be down). A payments service might need hourly backups and a one-hour restore; an internal wiki might accept daily backups and a day of downtime. Keep short-term backups in the same Region for fast restores and a copy in another Region or account for disasters.
Frequently asked questions
What should a container backup include?
Kubernetes resources (Deployments, Services, Secrets, CRDs), persistent volume data and the container images. Cloud resources outside the cluster, such as IAM roles and DNS, need their own backup or infrastructure as code.
Does AWS Backup support Amazon EKS?
Yes. AWS added native EKS support to AWS Backup in November 2025. It backs up cluster state and persistent data and can restore into a new cluster. Check AWS documentation for supported storage types and Regions.
Is Velero still relevant if my cloud has a native backup service?
Yes, when you need portability. Velero backups can be restored to a cluster on another cloud or on-premises, which native services do not aim to do. It is also the base for OADP and works with CloudCasa.
How do I back up container images?
Use registry replication: ECR cross-Region replication, ACR geo-replication, Artifact Registry multi-region repositories or Harbor replication. Pin images by digest so restores use identical builds.
Are etcd snapshots enough for Kubernetes backup?
No. etcd snapshots capture cluster state on self-managed control planes but not volume data or images. On managed services such as EKS, AKS and GKE you cannot access etcd directly, so use resource-level backups.
How often should I test restores?
At least quarterly and after major upgrades or tool changes. Restore into a separate cluster, confirm the application works and record the recovery time.
References
- AWS: AWS Backup now supports Amazon EKS (November 2025)
- AWS: Logically air-gapped vault support for Amazon EKS
- Microsoft Learn: Azure Kubernetes Service backup overview
- Google Cloud: Backup for GKE
- Red Hat: OpenShift documentation (OADP)
- Velero documentation
- Amazon ECR: Private image replication
- Microsoft Learn: ACR geo-replication
What's Your Reaction?
Like
0
Dislike
0
Love
0
Funny
0
Angry
0
Sad
0
Wow
0