Backup Principles and Solutions

Backup Principles

• Cluster instance

  A cluster instance consists of a config node, and multiple dds mongos and shard nodes. The config node is used to store the configuration information of a cluster instance, and the shard node is used to store data of a cluster instance. Backing up a cluster instance means that data on the config and shard nodes is backed up separately. The config and shard nodes in a cluster instance are backed up to their own hidden nodes. The backup process occupies certain CPU and memory resources of the hidden nodes. During the backup, the CPU usage, memory usage, and primary/standby delay of the hidden node increase slightly, which is normal. The backup files on the hidden nodes will then be compressed and stored in OBS, and the storage space of the instance will not be occupied.

• Replica set instance

  Replica set instance data is backed up on hidden nodes. The backup process occupies certain CPU and memory resources of the hidden node. During the backup, the CPU usage, memory usage, and primary/standby delay of the hidden node increase slightly, which is normal. The backup files on the hidden nodes will then be compressed and stored in OBS, and the storage space of the instance will not be occupied.

• Single node instance:

  Single-node instance backup is performed on only one node. The backup file is stored in OBS as a package, which does not occupy the storage of the instance.

  

  A single node instance is backed up using mongodump. During the backup, CPU and memory resources of the node are occupied. If the resources are insufficient, the backup fails. You are advised to migrate the single-node instance data to a replica set instance for backup.