Symptoms
- There is no difference or an increase in size between the original backup and a backup generated after a file is deleted.
- The ECS backup size is larger than the used disk space obtained from the file system.
Possible Causes
Possible causes are as follows:
- The backup mechanism itself causes this problem. The cloud server backups, cloud disk backups, and SFS Turbo backups created using CBR are all block-level backups. Different from file-level backups, block-level backups are performed by sector (512 bytes) each time.
- The metadata of the file systems on the disk occupies disk space.
- To reduce performance overhead, the file system adds a delete marker for the deleted file, but does not erase the data that has been written to the sector, and the metadata on the sector still exists. Block-level backups cannot detect whether data on a sector is deleted or not, but only determine whether a backup needs to be performed by checking whether all data blocks are zero blocks.
- CBR determines whether data in each sector changes by comparing two snapshots. Data changes include data addition, modification, and deletion. Backup is not performed if there are no data changes. If there are data changes, CBR further checks whether data blocks in the sector are all zero blocks. If so, backup is also not performed. Backups are performed only when there are non-zero blocks. If the data is deleted but metadata in the sector is not, the data block is also recognized as a non-zero block, and backups will be performed.