Overview
CPFS is designed to process and store large volumes of data with high throughput, high concurrency, and high availability. It is widely used in cloud computing, high-performance computing (HPC), big data analytics, media processing, and other scenarios that involve large datasets. This topic describes the CPFS backup feature of Cloud Backup, including backup methods, key benefits, how it works, the procedure, and billing.
Introduction
CPFS backup is a high-performance, cost-effective data protection solution that Alibaba Cloud Cloud Backup provides for your CPFS file systems. Cloud Backup supports CPFS for AI and CPFS General-purpose (including large-scale file systems), and provides advanced data protection capabilities such as concurrent multi-node scanning, storage API integration, deduplication, and compression. If files are accidentally deleted from a file system, you can restore the data from a Cloud Backup vault to keep your data safe.
Backup methods
CPFS backup provides two methods. Select a method based on your CPFS edition and region:
-
CPFS for AI (agentless backup): You only need to select a file system and a backup path in the console and associate a backup policy. No client installation and no additional compute resources are required. For more information, see Back up CPFS for AI.
-
CPFS General-purpose (agent-based backup): You must first install a backup client and add a data source, and then associate a backup policy. For more information, see Back up CPFS General-purpose.
-
Agentless backup for CPFS for AI is currently supported only in the China (Hangzhou) and China (Shanghai) regions. In regions where agentless backup for CPFS for AI is supported, new CPFS for AI backups use only the agentless method. Existing agent-based backup jobs for CPFS for AI have been recategorized as "CPFS General-purpose (agent-based backup)", and they continue to run without interruption.
Alibaba Cloud will contact you to assist with the upgrade to agentless backup. This upgrade allows you to release backup node resources and reduce costs. To upgrade sooner, submit a ticket.
-
In regions other than China (Hangzhou) and China (Shanghai), only agent-based backup for CPFS for AI is supported. For more information, see Back up CPFS.
Key benefits
-
Simple configurations
For CPFS for AI, you only need to select a file system and associate a backup policy in the console. No additional compute resources are required. For CPFS General-purpose, you need to deploy a lightweight backup client on an ECS instance as the compute node. Scheduled backup policies are intuitive and easy to configure.
-
Excellent backup performance
For CPFS for AI, the backup service reads and transfers data in parallel, so you do not need to plan the number of backup nodes. For CPFS General-purpose, multiple clients can back up one or more file systems in parallel, which significantly improves backup efficiency.
-
Source-side deduplication and compression
The data to be backed up is compressed and deduplicated, which improves backup performance and reduces network transfer costs and storage costs.
-
Backup lock to prevent accidental or malicious deletion
The backup lock feature prevents backup data in a vault from being deleted by any account or method before the configured retention period expires.
-
Cross-region or cross-account backup
Cross-account backup and cross-region, cross-account backup vault replication are supported, providing multi-layered disaster recovery protection across regions and accounts.
How it works
-
CPFS for AI: Agentless backup is used. You do not need to deploy a backup client or prepare additional compute resources. After you select a file system and a backup path in the console and associate a backup policy, the service scans the files under the path, performs data reduction such as deduplication and compression, and then writes the incremental data to the backup vault.
-
CPFS General-purpose: First, install and activate a Cloud Backup client on an ECS instance, and then log on to the Cloud Backup console to add the CPFS data source. During a backup job, the client scans the specified folders in the file system, identifies the files to be backed up, performs data reduction such as deduplication and compression, and then uploads the incremental data to the backup vault. If you install multiple clients, CPFS data can be backed up in parallel to improve backup performance.
-
Cloud Backup uses a forever incremental backup mechanism. Except for the first full backup, each subsequent backup uploads only the data that has changed since the previous backup. A complete full backup is synthesized in the cloud to minimize network and storage consumption.
-
When you delete any backup, only the data segments unique to that backup are deleted. Other backups are not affected.
-
Deleting a full backup does not affect subsequent data restoration. This is because deleting a full backup only removes the data segments exclusive to that backup. Data segments referenced by other backups are not deleted.
-
Deleting an intermediate backup does not affect the incremental backups or the recovery after that backup. This is because file backup is forever incremental: the first backup is a full backup, and each subsequent backup uploads only the data that has changed since the previous backup.
-
-
In scenarios where backups of on-premises large-scale file systems are sent to the cloud, Cloud Backup supports connections to Alibaba Cloud over the Internet, Express Connect, or a VPN. For backups over the Internet, the server on which the client is installed only needs to access the Internet. You do not need to expose the IP address of the server to the Internet.
Procedure
CPFS for AI
-
Activating Cloud Backup is free of charge. You are charged only for storage capacity.
-
On the CPFS Backup page, click the CPFS for AI tab. Select a file system, configure the backup path and filter rules, and associate a backup policy. The first time you use this feature, you must authorize a service-linked role.
-
If files in a file system are lost or damaged, you can restore all files or specific files that meet your conditions from historical recovery points to the source file system or a new file system.
CPFS General-purpose
-
Activating Cloud Backup is free of charge. You are charged only for storage capacity. For more information about billing, see Billing methods and billable items.
-
The backup client establishes communication and management services between the backup node (ECS) and Cloud Backup.
-
Add CPFS as a data source in the Cloud Backup console.
-
Select the target data source, configure the backup file path and filter rules, and associate a backup policy to create a backup plan. Cloud Backup then starts the plan and continuously backs up the file system as scheduled.
Important-
After a backup plan is created, the first backup job is a full backup. Subsequent jobs are incremental backups by default.
-
If your selected region supports backup policies, you can create a backup plan only by associating it with a backup policy. Cloud Backup then periodically backs up the file system according to that policy.
-
In the left-side navigation pane of the console, click Policy Center to view the regions that support backup policies. For information about how to create a backup policy, see Create a backup policy.
-
-
If files in a file system are lost or damaged, you can restore all files or specific files that meet your conditions from historical recovery points to the source CPFS file system or a new CPFS file system.
Billing
CPFS backup does not support the 30-day free trial.
CPFS backup mainly incurs storage capacity fees, which are calculated based on the vault capacity that your backup data actually consumes. You can view the details of the backup vault on the Overview page of the Cloud Backup console.
In addition, Cloud Backup charges other fees for specific backup configurations or operations. For example, CPFS for AI supports recovery point virus scan. After you enable this feature, you are charged based on the number of files scanned. For more information, see Billing methods and billable items.
For the prices of billable items, see Pricing.