AnalyticDB for PostgreSQL provides base backup and log backup on a regular basis to prevent data loss.
How it works
AnalyticDB for PostgreSQL uses base backup and log backup to restore instance data to a specific point in time and keep the state and data of a distributed instance consistent.
Backup type | Elastic storage mode | Serverless mode |
Base backup | Base backup backs up all data of an instance. Base backup compresses a snapshot of all instance data and stores the snapshot to offline storage media. Base backup does not block your read or write operations. The operation logs that are generated during a base backup are also backed up to ensure the integrity of base backup.
| Base backup backs up the metadata of a database. The Serverless mode uses a storage-compute-decoupled architecture. Data is stored in Object Storage Service (OSS) and does not require backup. Instance metadata is compressed and stored in the same OSS directory. Base backup does not block your read or write operations. |
Log backup (also called incremental backup) | Log backup backs up log files that are generated from an instance and stores them to offline storage media. Log files record the DML and DDL operations that are performed on databases. | |
AnalyticDB for PostgreSQL uses a full base backup and continuous log backups to restore data to new instances to a specific point in time and ensure data security.
After the base backup data is restored, all nodes execute the data change statements that are recorded in the log backup files in sequence until the point in time for restoration. The point in time ensures consistency among all nodes when the data restoration of each node is complete.
Backup policies
Elastic storage mode | Serverless mode |
View and modify backup policies. For more information, see the "View and modify backup policies" section of the Back up data topic. | The system performs incremental backup for data that is added or removed within a backup cycle without compressing the data. The default backup cycle is seven days, and the ratio of backup to data storage is about 68%. For more information, see the "Backup policy" section of the Backup and restoration (Serverless mode) topic. To change the retention period of the backup data, Submit a ticket. |
Scenarios
The backup and restoration feature can be used in the following scenarios:
Instance data is lost for reasons such as accidental changes.
The source instance data is corrupted or unavailable.
An instance needs to be cloned with the same data.
The instance specifications except for the number of nodes need to be changed.
Limits
Data can be restored from the source instance only when the source instance has at least one backup set in the successful state.
Billing
Elastic storage mode | Serverless mode |
The backup and restoration feature is available free of charge during public preview. You may be charged for this feature in the future. | The backup and restoration feature is not billed separately. Backup data uses disk storage and may incur storage fees. For more information about the billing methods and pricing of disk storage, see Pricing. |
References
Elastic storage mode
Serverless mode