Data protection and disaster recovery in IBM Spectrum Scale

The IBM Spectrum Scale installation should be protected against data loss to ensure continuity of operation after a malfunction.

Data loss can be prevented by protecting the four types of key data:
  • Cluster configuration data
  • File system configuration data
  • File system contents (user data, metadata, configuration)
  • Protocol configuration data
Cluster configuration data is the administrative data that associates the nodes, addresses, networks, and software installation on each node. The system administrators should save the following configuration information:
  • The output of the mmlscluster command to ensure that reconstruction of this data is possible if needed.
  • CCR backup file or the mmsdrfs file depending on type of repository. For more information, see mmsdrrestore command.
  • Snapshots

File system configuration data consists of a wide variety of information involving all the file systems in the cluster. To protect this data, it is essential to use the mmbackupconfig command for each file system and save the output files for future use. This configuration data details which disks are associated as NSD components of which file systems, how much storage is in use, the filesets defined, the quotas defined, and other valuable configuration data that describes the file system structure. These do not include the file data in the user files. User file data is the lowest level of information and most frequently changing contents that need protection.

If you want to back up protocols configuration information, see Backing up and restoring protocols and CES configuration information.

IBM Spectrum Scale has built in data protection processes that allows the users not only to back up and restore valuable data, but also recover data that could be potentially lost or corrupted by their own actions. For example, unintentional deletion or overwriting of a user's file.