ApsaraDB for HBase Performance-enhanced Edition provides a cost-effective and efficient backup and restoration feature. This topic describes how to enable it.
Benefits
Achieve a recovery point objective (RPO) of less than 30 seconds with real-time incremental data synchronization.
NoteAn RPO is the maximum acceptable amount of data loss after an application failure. For example, an RPO of 15 minutes means that if a failure occurs, the data from the last 15 minutes may not be recoverable.
Performs full database backups by default and supports restoration of individual tables.
Supports point-in-time recovery.
Limitations
This feature is available only for ApsaraDB for HBase Performance-enhanced Edition clusters.
Backup and restoration of Solr data is not supported.
If storage space is insufficient, the backup process is interrupted, but existing backup data is retained.
Billing
Fees for the backup and restoration feature include cold storage fees and tunnel service fees.
Cold storage fees vary by region. For details, see the following tables.
Region
Fee
Chinese mainland
Subscription: CNY 0.2/GB
Pay-as-you-go: CNY 0.00042/GB
China (Hong Kong)
Subscription: CNY 0.23/GB
Pay-as-you-go: CNY 0.00048/GB
Germany (Frankfurt)
Subscription: CNY 0.45/GB
Pay-as-you-go: CNY 0.00094/GB
The tunnel service is billed based on the number of core nodes.
Enable backup and restoration
Log in to the ApsaraDB for HBase console.
In the upper-left corner of the top navigation bar, select the region where your cluster is located.
On the Clusters page, click the ID of the target cluster.
In the left-side navigation pane, click Backup and Restoration.
Click Enable Now. The Enable backup recovery dialog box appears.
Specify the Number of Core nodes and Estimated cold storage capacity (GB). For more information, see Cluster specifications.
Click OK.
Configure automatic backups
Log in to the ApsaraDB for HBase console.
In the upper-left corner of the top navigation bar, select the region where your cluster is located.
On the Clusters page, click the ID of the target cluster.
In the left-side navigation pane, click Backup and Restoration.
Click Backup Configuration.
In the Backup Configuration dialog box, configure the backup parameters based on your business requirements.
Parameter
Description
Tables to back up
You can use an asterisk (*) to back up all tables in the database.
Full backup interval (days)
By default, a full backup runs every 7 days. You can set the interval from 3 to 10 days. A short interval may cause backups to fail, while a long interval can increase restoration time.
Next full backup time
Specifies the start time for the next full backup. Schedule full backups during off-peak hours.
Number of retained full backups
The number of full backups to retain.
Click OK.
Restore cluster data
Log in to the ApsaraDB for HBase console.
In the upper-left corner of the top navigation bar, select the region where your cluster is located.
On the Clusters page, click the ID of the target cluster.
In the left-side navigation pane, click Backup and Restoration.
Click Initiate data recovery.
In the Recovery dialog box, configure the restoration parameters based on your business requirements.
Parameter
Description
Target cluster
Select the cluster to which you want to restore data.
Point in time
Select the point in time to restore to.
Full database restoration
Select Yes to restore the entire database.
Tables to restore
If you select No for No, you must specify the tables to restore. Asterisks (*) are not supported.
To restore a table, use the format:
namespace:table. Example:default:testTable.To restore a table to another table, use the format:
namespace:table/namespace:table2. Example:default:testTable/default:testTable2.
Click OK.
Storage space estimation
Storage space for backup and restoration = Total size of full backups + Total size of incremental backups.
Total size of full backups = (Number of retained full backups + 1) × Size of a single full backup.
The "+1" is included because the system deletes a previous backup only after the new one is successfully created. This process temporarily requires storage for one additional full backup.
You can find the size of a full backup on the HBase Master UI or the cluster management page.
Total size of incremental backups = Log retention period in days × Daily incremental log size.
For example, with an average write speed of 1 MB/s, you would need 84 GB of space per day. You can check the cluster's write speed on the monitoring page or contact Alibaba Cloud technical support for an assessment. For write-heavy clusters, the daily log growth can be significant. You can reduce the storage required for logs by decreasing the full backup interval.
Ensure you properly estimate the required storage space. Insufficient space causes backup tasks to fail.
FAQ
Q: What is the most recent point in time to which I can restore data after a failure?
A: Under normal conditions with no backlogs, the recovery point depends on how frequently HBase persists logs to Object Storage Service (OSS). By default, the system flushes logs every 30 seconds, resulting in a maximum data loss of 30 seconds.
Q: How long does a restoration take?
A: The time required depends on the data volume and bandwidth limits.
Full data restoration speed: The maximum bandwidth from OSS is 1 GB/s, and a single Lindorm Tunnel Service (LTS) node provides 100 MB/s.
Incremental data restoration speed: The target ApsaraDB for HBase cluster can write at 30 to 40 MB/s per node, while a single LTS node provides 100 MB/s.
Q: Can I restore a single table?
A: Yes. If a table with the same name already exists, you can restore the backup data to a new table with a different name.