File system and Storage Checkpoint space management
To support BLI backups, the VxFS file systems need extra disk space to keep track of the block change information. The space that is required depends on the type of checkpoint that is used and the database change rate while the backup is running.
Using Storage Checkpoints has an effect on space in the following ways:
If the database is offline during the entire backup window (a cold database backup) or you use this checkpoint type, the additional space is minimal. Each file system requires about 1% of free space. This checkpoint sets a bit to indicate if a file block changed. When you use this checkpoint type, the data files are left in quiesce (write suspend) mode for the duration of the backup. | |
If the database is online during the backup and using this checkpoint type, then more free space is needed in the file system. NetBackup for Oracle keeps the Oracle containers in quiesce (write suspend) mode only for the time that is needed to create a Storage Checkpoint. During the backup, the checkpoint creates copies of any file blocks immediately before they are changed. The backup up contains only the unchanged blocks and the original copies of the changed blocks. After the backup completes, the Fulldata Storage Checkpoint is converted to a Nodata Storage checkpoint and the copied blocks are returned to the free list. If the workload change rate is light during backup or the backup window is short, 10% free space is usually sufficient for the workload. If the database has a heavy change rate while the backup is running, the file systems may require more than 10% of free space. |
Note:
The default option that NetBackup uses for backups is Fulldata Storage Checkpoint.
To use Nodata Storage Checkpoint instead of the default option, a user must create the following empty touch file:
/usr/openv/netbackup/ext/db_ext/NODATA_CKPT_PROXY