The backup center of Container Service for Kubernetes (ACK) provides an end-to-end solution for backing up, restoring, and migrating stateful and stateless applications. The backup center also provides disaster recovery and application migration capabilities for stateful applications deployed in hybrid cloud and multi-cluster environments. For example, you can use the backup center to periodically back up the data and applications in an ACK cluster, restore them from backups at a time, and migrate them across zones or regions. This topic describes how the backup center works and the usage scenarios.
How it works
The backup center uses Velero to back up and restore applications (resource YAML files) in ACK clusters and stores backup files in the specified Object Storage Service (OSS) buckets.
The backup center uses the disk snapshot feature of Elastic Compute Service (ECS) to back up and restore disk volumes. For more information, see Create a snapshot for a disk and Create a disk from a snapshot.
The backup center uses Cloud Backup to back up and restore volumes that use local disks, Apsara File Storage NAS (NAS) file systems, OSS buckets, CPFS file systems, and open source off-tree drivers (such as NFS and Ceph). The backup center also allows you to back up and restore volumes in hybrid cloud scenarios. For more information, see What is Cloud Backup?.
Cross-region data synchronization supports only disk volumes. You can create disk snapshots in Region A, and replicate the disk snapshots to Region B and then restore data from the snapshots.
When you restore volumes from backups created by using Cloud Backup, you can use StorageClass conversion to specify the StorageClass of the restored volumes. The CSI provisioner of the ACK cluster will create and mount persistent volumes (PVs) of the specified StorageClass. Cloud Backup then restores data to the corresponding storage, such as disks and NAS file systems.
ImportantWhen the backup center creates NAS, OSS, and local disk volumes, the backup center uses Cloud Backup to back up files. Cloud Backup occupies NAS and OSS storage bandwidth. We recommend that you create backups during off-peak hours. For more information, see Cloud Backup.
Use scenarios
Scenario 1: Application disaster recovery and migration
Disaster recovery of applications that use PVs in an ACK cluster and migration of applications across clusters that are deployed in the same region.
The backup center can back up and restore applications and data within ACK clusters and across ACK clusters that are deployed in the same region. After you create backups in the backup cluster, you can use StorageClass conversion to convert the StorageClass of the volumes to be restored or directly restore applications in the restore cluster.
The backup center provides the following features for data stored on the cloud:
Create fast snapshots for full data at a point in time.
Periodically back up incremental data.
Create fast snapshots for TB-level disks.
The following cloud storage types are supported:
Local storage: local disks.
On-cloud block storage: disks.
On-cloud file storage: NAS and CPFS.
On-cloud object storage: OSS.
Application disaster recovery and restoration across regions on the public cloud
You can back up and restore applications and data across ACK clusters that are deployed in different regions. After you create backups in the backup cluster, you can directly restore data in the restore cluster.
Create fast snapshots for full data at a point in time.
Periodically back up incremental data.
Create fast snapshots for TB-level disks.
The following cloud storage types are supported:
On-cloud block storage: disks.
Application disaster recovery and migration in hybrid cloud environments
After you connect Kubernetes clusters in data centers or Kubernetes clusters on other cloud service platforms to registered clusters, you can use the backup center to periodically back up applications and data to the cloud and restore them in the specified ACK clusters. For more information about how to connect Kubernetes clusters to registered clusters, see Clusters. The backup center provides the following features for disaster recovery in hybrid cloud environments:
Create fast snapshots for full data at a point in time.
Periodically back up incremental data.
Perform incremental data minimization, data deduplication, and data compression.
The following on-premises storage types are supported:
On-premises block storage such as Storage Area Network (SAN), Ceph OBD, and Logical Volume Manager (LVM).
On-premises file systems such as HostPath, NFS, NetApp, GlusterFS, and CephFS.
On-premises object storage such as Ceph S3.
Scenario 2: Application disaster recovery (resource YAML file backup and restoration)
You can use the backup center to back up and restore applications, including the YAML files of persistent volume claims (PVCs) and PVs, across ACK clusters within the same region, ACK clusters in different regions, and registered clusters. If you do not exclude PVCs and PVs when you back up or restore applications, Velero restores all PVCs and PVs used by the applications and associates them with the original data sources (with the same disk IDs or NAS mount targets).
Applications can share data stored in shared volumes, such as NAS and OSS. Therefore, you need to ensure the consistency of concurrent reads and writes. For volumes that cannot be shared, such as disk volumes, you need to first unmount the volumes. Otherwise, exceptions such as resource contention or restoration failures may occur.
You must install the corresponding volume plug-in in the restore cluster and make sure that the restore cluster meets the mounting requirements. For example, you cannot mount disks across zones and you can access NAS file system only through internal networks.
Scenario 3: PVC-to-PVC volume backup and restoration
You can use the backup center to back up and restore disks within an ACK cluster, across ACK clusters within the same region, and across ACK clusters in different regions. After you create backups in the backup cluster, you can use the VolumeSnapshot resource in the cluster to launch a new application, or use the backup center to create new disk PVs and PVCs from backups in another cluster.
Create fast snapshots for full data at a point in time.
Periodically back up incremental data.
Create fast snapshots for TB-level disks.
To avoid resource contention, you can add
prefixesorsuffixesto the names of PVCs to be restored.
The following cloud storage types are supported:
On-cloud block storage: disks.
Terms
Term | Description |
Crash-consistent backup | Crash-consistent backup does not guarantee that the application cache and operating system cache are backed up in the storage. |
Application-consistent backup | Application-consistent backup guarantees that the application cache and operating system cache are backed up in the storage. |
Cross zones | Application backup and restoration across zones. |
Cross regions | Application backup and restoration across regions. |
Snapshot | Snapshots of file systems and disks. |
Fast snapshot | Snapshots of TB-level disks created within a few seconds. |
Restoration | Restoration of applications and data. |
Application backup | Backup of metadata of applications, services, and configurations. |