Introduction to Backups

What Are Database Backups

Automated backups are created during the backup window of your TaurusDB instances. The backups are stored based on a preset retention period (1 to 732 days).

A backup is generated each time a backup is complete. If the instance fails or data is damaged, you can use the backup to restore the instance, ensuring data availability.

Backup Types

TaurusDB supports multiple backup types. For details, see Backup Types.

Backup Principles

TaurusDB uses Huawei's DFV storage, which decouples storage from compute.

• Compute layer: provides services for external systems and manages logs.
• Storage layer: consists of Common Log nodes and Slice Store nodes. Common Log nodes store logs. Slice Store nodes store data.

As shown in the following figure, the creation of backups involves both the compute and storage layers.

• The primary node at the compute layer reads the logs of the Common Log nodes at the storage layer and backs them up to OBS.
• The primary node at the compute layer delivers a command for backing up data to the Slice Store nodes at the storage layer. The Slice Store nodes back up data to OBS.

During the creation of a backup, the CPU usage and memory usage of the primary node of your instance increase slightly, but you will not notice anything at the storage layer. The final backup is stored in OBS as multiple data files, so it does not use any storage on the instance.

Figure 1 Backup principles

How Data Is Backed Up

TaurusDB automated backups are enabled by default and cannot be disabled. TaurusDB performs automated full backups based on the time window and cycle specified in the backup policy. It also backs up data changes made after the most recent automated full backup or incremental backup every 5 minutes or when a certain amount of incremental data is generated.

• Restoration from backups: The system restores instance data to the state at the end of the full backup you choose. This action restores the data for the entire instance.
• Point-in-time recovery (PITR): When you restore an instance to a point in time, the system downloads the most recent full backup from OBS and uses it for restoration. After the restoration is complete, the system replays incremental backups to the specified point in time.

Figure 2 How data is backed up

Backup Storage and Billing

TaurusDB backups are stored on OBS. They do not occupy the storage of your instance. TaurusDB provides free backup space of the same size as your purchased storage.

If your instance is frozen, no free backup space is available. As a result, the original automated backups of the instance will be billed.

• If you unfreeze the instance, the free backup space will be restored.
• If you directly delete the frozen instance, its automated backups will also be deleted and the backup space will not be billed any longer.

Example:

If you purchase a DB instance with 200 GB of storage, you can get an additional 200 GB of backup space and will only be billed for backups in excess of 200 GB. The first 200 GB of backup data is free. After the free backup space is used up, the additional space used will be billed on a pay-per-use basis. For pricing details, see Product Pricing Details.

Backup Clearing