ApsaraDB for HBase:Enable backup and recovery

Limitations

• Data backup and restoration is available only for ApsaraDB for HBase Performance-enhanced Edition.

• If storage space runs out, ongoing backups are interrupted. Data already backed up is retained.

Billing

Enabling data backup and restoration incurs two types of charges: cold storage fees and Lindorm Tunnel Service (LTS) fees.

Cold storage fees by region:

LTS fees vary based on the number of core nodes. For details, see Cluster specifications.

Enable data backup and restoration

1. Log on to the ApsaraDB for HBase console.

2. In the top navigation bar, select the region where your cluster is deployed.

3. On the Clusters page, click the cluster ID.

4. In the left-side navigation pane, click Backup and Restoration.

5. Click Enable Now. The Enable backup recovery dialog box appears.

6. Set Number of Core nodes and Estimated Cold Storage Capacity.

7. Click OK.

Configure automatic backups

1. Log on to the ApsaraDB for HBase console.

2. In the top navigation bar, select the region where your cluster is deployed.

3. On the Clusters page, click the cluster ID.

4. In the left-side navigation pane, click Backup and Restoration.

5. Click Backup Configuration.

6. In the Backup Configuration dialog box, configure the following parameters.

   Parameter	Description
   Backup table	Specify the tables to back up. Use an asterisk (*) to back up all tables in the database.
   Full backup cycle (days)	How often full backups run. Default: 7 days. Range: 3–10 days. A shorter interval increases the chance that a backup job will not finish within the cycle. A longer interval increases restoration time.
   Next full backup time	The time the next full backup starts. Schedule this during off-peak hours to minimize performance impact.
   Number of full backup reservations	How many full backups to retain.

7. Click OK.

Restore data

1. Log on to the ApsaraDB for HBase console.

2. In the top navigation bar, select the region where your cluster is deployed.

3. On the Clusters page, click the cluster ID.

4. In the left-side navigation pane, click Backup and Restoration.

5. Click Initiate data recovery.

6. In the Recovery dialog box, configure the following parameters.

   Parameter	Description
   Recovery cluster	The name of the cluster to restore data to.
   Time point	The point in time to restore to.
   Full database recovery	Select Yes to restore all tables. Select No to restore specific tables.
   Restore table	Available when Full database recovery is set to No. Asterisks (*) are not supported. Use one of the following formats:<br>- namespace:table — restores a table in place (for example, default:testTable).<br>- namespace:table/namespace:table2 — restores data from one table to another (for example, default:testTable/default:testTable2).

7. Click OK.

Estimate storage requirements

Total storage = storage for full backups + storage for incremental backups.

Full backup storage:

Full backup storage = (number of retained full backups + 1) × size of one full backup

The +1 accounts for the previous full backup, which is retained until a new full backup completes.

Check your current full backup size on the Cluster Management page.

Incremental backup storage:

Incremental backup storage = number of log retention days × daily incremental log size

As a reference, a cluster writing at an average speed of 1 MB/s generates approximately 84 GB of incremental data per day. Check your cluster's write speed on the Monitoring and Alerts page. If your cluster ingests large data volumes daily, shorten the full backup interval to reduce the size of incremental logs that need to be retained.

Plan your storage capacity in advance. Backups are interrupted when storage space is exhausted.

FAQ

What is the most recent point in time I can restore to after a failure?

Logs are uploaded to Object Storage Service (OSS) once every 30 seconds or once every 30 MB, whichever comes first. After a failure, data written in the last 30 seconds before the failure may not be recoverable.

How long does a restoration take?

Restoration time depends on data size and available bandwidth.

• Full data restoration: OSS delivers data at up to 1 GB/s. Each LTS node writes at up to 100 MB/s.

• Incremental data restoration: Each cluster node can receive 30–40 MB/s of data. Each LTS node writes at up to 100 MB/s.

Can I restore a single table?

Yes. If a database has multiple tables with the same name, you can also restore data from one table to another by specifying the source and destination in the namespace:table/namespace:table2 format.