The Container Service for Kubernetes (ACK) Backup Center is a one-stop solution for backing up, recovering, and migrating stateless and stateful applications. It provides disaster recovery and application migration for stateful applications, particularly in hybrid cloud and multi-cluster environments. You can perform scheduled backups, recover applications and data within a cluster, and migrate them across zones and regions. This topic describes how the Backup Center works, its billing, and its scenarios.
How it works
It uses Velero to back up and recover applications (resource YAMLs) in a cluster. The backup data is securely stored in a specified Object Storage Service (OSS) bucket.
It uses the snapshot feature of ECS to back up and recover data from cloud disk volumes. For more information, see Create a snapshot and Create a data disk from a snapshot.
It uses Cloud Backup to back up and recover data from other types of volumes or in hybrid cloud scenarios. These volume types include local disks, Alibaba Cloud NAS, OSS, CPFS, and volumes that use open-source out-of-tree drivers, such as NFS and Ceph. For more information, see What is Cloud Backup?.
Cross-region data synchronization currently supports only cloud disk volumes. It works by creating a cloud disk snapshot in Region A and copying it to Region B for data restoration.
When you recover data using Cloud Backup, you can transform the storage class to restore the data to a specified storage type. The cluster's CSI provisioner creates and attaches a persistent volume (PV) based on the target storage class. Then, Cloud Backup recovers the data to the corresponding underlying storage, such as a cloud disk or NAS.
ImportantWhen you back up NAS, OSS, or local disk volumes, the Backup Center uses Cloud Backup to perform a file backup. This operation consumes NAS and OSS storage bandwidth. Perform backups during off-peak hours. For more information, see What is Cloud Backup?.
Billing
The Backup Center feature is free of charge. However, you are charged for the storage that is used by related services.
Object Storage Service (OSS) fees: An OSS bucket stores the backed-up cluster resources (YAML files).
Cloud disk snapshot fees: Snapshots are used to back up data from cloud disk volumes.
NoteStarting from 11:00 on October 12, 2023, Alibaba Cloud no longer charges storage fees or feature fees for the snapshot instant access feature of ECS cloud disks. For more information, see Snapshot instant access.
When you use ESSD PL0, PL1, PL2, PL3, or ESSD AutoPL disks, the snapshot instant access feature is enabled by default for snapshots created during backup.
Cloud Backup fees: Cloud Backup is used to back up data from volume types other than cloud disks.
Scenarios
Scenario 1: Overall disaster recovery and migration for application data
Disaster recovery for persistent applications and migration between clusters in the same region
You can back up and recover applications and data within an ACK cluster or between ACK clusters in the same region. After you complete the backup in the source cluster, you can change the underlying storage type by transforming the storage class in the recovery cluster, or you can recover the data directly.
For data in the cloud, the Backup Center provides the following features:
Point-in-time full instant snapshot backups.
Scheduled incremental backups.
Instant snapshots for terabyte-scale cloud disks.
The following cloud storage types are supported:
Local storage: local disks.
Cloud block storage: cloud disks.
Cloud file storage: NAS, CPFS, and others.
Cloud object storage: OSS.
Disaster recovery and application migration between clusters across regions
You can back up and recover applications and data between ACK clusters in different regions. After you complete the backup in the source cluster, you can recover the data directly in the recovery cluster.
Point-in-time full instant snapshot backups.
Scheduled incremental backups.
Instant snapshots for terabyte-scale cloud disks.
The following cloud storage types are supported:
Cloud block storage: cloud disks.
Disaster recovery and application migration in a hybrid cloud scenario
You can connect a self-managed Kubernetes cluster in an on-premises data center or a Kubernetes cluster from another cloud provider to a registered cluster. Then, you can use the Backup Center to schedule backups of applications and data to the cloud. You can then recover the entire application and its data in a target ACK cluster. For more information about how to connect a Kubernetes cluster to a registered cluster, see Create a registered cluster. For data backups, the hybrid cloud disaster recovery feature of the Backup Center provides the following capabilities:
Point-in-time full snapshot backups.
Scheduled incremental backups.
Minimized data increments, deduplication, and compression.
The following on-premises storage types are supported:
On-premises block storage: SAN, Ceph RBD, and LVM for local disks.
On-premises file systems: HostPath, NFS, NetApp, GlusterFS, and CephFS.
On-premises object storage: Ceph S3.
Scenario 2: Application disaster recovery by backing up and recovering cluster resource YAMLs
You can back up and recover applications between ACK clusters or registered clusters in the same region or across different regions. This includes the YAML files for persistent volume claims (PVCs) and PVs that the application uses. If you do not exclude PVCs and PVs during backup and recovery, Velero restores them along with other resources. The restored resources point to the original data source, such as a cloud disk ID or a NAS mount target.
For storage types that support shared mounts, such as NAS and OSS, both applications share the data. Your business logic must ensure data consistency during concurrent reads and writes. For storage types that do not support shared mounts, such as cloud disks, you must manually detach the cloud disk first. Otherwise, unexpected issues such as resource contention or recovery failures may occur.
The new recovery cluster must have the corresponding storage plugin installed. You must also confirm that the requirements for mounting the underlying storage are met. For example, a cloud disk instance cannot be mounted across zones, and NAS supports access only over an internal network.
Scenario 3: PVC-to-PVC volume backup and recovery
You can back up and recover cloud disk data within an ACK cluster, or between ACK clusters in the same region or across different regions. After you complete the backup in the source cluster, you can use the VolumeSnapshot resource created in the cluster to start a new application directly. You can also use the Backup Center to recover a new set of cloud disk PVCs and their corresponding PVs in another cluster for a new application to use.
Point-in-time full instant snapshot backups.
Scheduled incremental backups.
Instant snapshots for terabyte-scale cloud disks.
Supports adding a
prefixorsuffixto the names of recovered PVCs to avoid resource contention.
The following cloud storage types are supported:
Cloud block storage: cloud disks.
Core concepts
Concept | Description |
Crash-consistent backup | Does not guarantee that the application cache and operating system cache are fully backed up to the storage medium. |
Application-consistent backup | Guarantees that the application cache and operating system cache are backed up to the storage medium. |
Cross-zone | Backup and recovery across different zones in the same region. |
Cross-region | Backup and recovery across different regions. |
Snapshot | File system snapshots and block storage snapshots for cloud disks. |
Instant snapshot | Second-level snapshot creation for terabyte-scale cloud disks. |
Recovery | Unified restoration of applications and data. |
Application backup | Metadata backup for applications, services, and configurations. |
Data protection | Backup of volume resources and their associated underlying storage data. |